1 Name: Curating the Clinical Genome 2018 Wellcome Genome Campus Conference Centre, Hinxton, Cambridge, UK 23-25 May 2018 Scientific Programme Committee: Helen Firth Cambridge University Hospitals, UK Gert Matthijs University of Leuven, Belgium Heidi Rehm Harvard Medical School, USA Marc Williams Geisinger Health System, USA Caroline Wright University of Exeter, UK Tweet about it: #CCG18 @ACSCevents /ACSCevents /c/WellcomeGenomeCampusCoursesandConferences
Transcript
1
Name:
Curating the Clinical Genome 2018
Wellcome Genome Campus Conference Centre, Hinxton, Cambridge, UK
23-25 May 2018
Scientific Programme Committee:
Helen Firth Cambridge University Hospitals, UK Gert Matthijs University of Leuven, Belgium Heidi Rehm Harvard Medical School, USA Marc Williams Geisinger Health System, USA Caroline Wright University of Exeter, UK
Conference Badges Please wear your name badge at all times to promote networking and to assist staff in identifying you. Scientific Session Protocol Photography, audio or video recording of the scientific sessions, including poster session is not permitted. Social Media Policy To encourage the open communication of science, we would like to support the use of social media at this year’s conference. Please use the conference hashtag #CCG18. You will be notified at the start of a talk if a speaker does not wish their talk to be open. For posters, please check with the presenter to obtain permission. Internet Access Wifi access instructions:
Join the ‘ConferenceGuest’ network
Enter your name and email address to register
Click ‘continue’ – this will provide a few minutes of wifi access and send an email to the registered email address
Open the registration email, follow the link ‘click here’ and confirm the address is valid
Enjoy seven days’ free internet access!
Repeat these steps on up to 5 devices to link them to your registered email address Presentations Please provide an electronic copy of your talk to a member of the AV team who will be based in the meeting room. Poster Sessions Posters will be displayed throughout the conference. Please display your poster in the Conference Centre on arrival. There will be two poster sessions during the conference. Odd number poster assignments will be presenting in poster session 1, which takes place on Wednesday, 23 May at 17:45-19:15. Even number poster assignments will be presenting in poster session 2, which takes place on Thursday, 24 May at 18:15-19:45. The abstract page number indicates your assigned poster board number. An index of poster numbers appears in the back of this book. Conference Meals Lunch and dinner will be served in the Hall, apart from lunch on Wednesday, 23 May when it will be served in the Conference Centre, and on Friday, 25 May when it will be a take away lunch. Please refer to the conference programme in this book as times will vary based on the daily scientific presentations. Please note there will be no lunch or dinner facilities available outside of the conference timetable.
Please inform the conference organiser if you are unable to attend the dinners.
5
Dietary Requirements If you have advised us of any dietary requirements, you will find a coloured dot on your badge. Please make yourself known to the catering team and they will assist you with your meal request. If you have a gluten or nut allergy, we are unable to guarantee the non-presence of gluten or nuts in dishes even if they are not used as a direct ingredient. This is due to gluten and nut ingredients being used in the kitchen. Social Events The Hall Bar (cash bar) will be open during dinner each day. The Conference Centre Forum Bar (cash bar) will be open after dinner each day. Wednesday, 23 May – A drinks reception will take place during the poster session in the Conference Centre Forum from 17:45 followed by dinner at 19:15. Thursday, 24 May – A drinks reception will take place during the poster session in the Conference Centre Forum from 18:15 followed by dinner at 19:45. All conference meals and social events are for registered delegates. For Wellcome Genome Campus Conference Centre Guests Check in If you are staying on site at the Wellcome Genome Campus Conference Centre, you may check into your room from 14:00. The Conference Centre reception is open 24 hours. Breakfast Your breakfast will be served in the Hall restaurant from 07:30 – 09:00 Telephone If you are staying on-site and would like to use the telephone in your room, you will need to contact the Reception desk (ext. 5000) to have your phone line activated. They will require your credit card number and expiry date to do so. Departures You must vacate your room by 10:00 on the day of your departure. Please ask at reception for assistance with luggage storage in the Conference Centre. Taxis Please find a list of local taxi numbers below: All Journeys Cam-Air-Connect (Airport Specialist) [email protected] +44 (0)1223 750850 Sawston Cab Co Ltd (Airport Specialist) [email protected] +44 (0)1223 517008 For Cambridge & the airports Panther Taxis www.panthertaxis.co.uk +44 (0)1223 715715
For Audley End & Great Chesterford Station Walden Cabs +44 (0)1799 500500 Crocus +44 (0)1799 525511 For Whittlesford Station & The Holiday Inn Express Mid Anglia +44 (0)1223 836000 Caz Cars +44 (0)1223 513693 Return Ground Transport Complimentary return transport has been arranged for 13:15 on Friday, 25 May to Cambridge station and city centre (Downing Street, Cambridge), and Stansted and Heathrow airports. A sign-up sheet will be available at the conference registration desk from 15:00 on Wednesday, 23 May. Places are limited so you are advised to book early. Please allow a 30 to 40 minute journey time to both Cambridge and Stansted Airport, and 2.5 to 3 hours to Heathrow due to possible traffic delays. Messages and Miscellaneous Lockers are located outside the Conference Centre toilets and are free of charge. All messages will be available for collection from the registration desk in the Conference Centre. A number of toiletry and stationery items are available for purchase at the Conference Centre reception. Cards for our self-service laundry are also available. Certificate of Attendance A certificate of attendance can be provided. Please request one from the conference organiser based at the registration desk. Contact numbers Wellcome Genome Campus Conference Centre – 01223 495000 (or Ext. 5000) Wellcome Genome Campus Conference Organiser (Zoey) – 07747 024256 If you have any queries or comments, please do not hesitate to contact a member of staff who will be pleased to help you.
7
Conference Summary Wednesday, 23 May 2018 12:00 – 13:00 Registration with lunch 13:00 – 13:15 Welcome and introduction 13:15 – 15:00 Session 1: Data sharing 15:00 – 15:30 Afternoon tea 15:30 – 17:15 Session 2: Variant guidelines and resources 17:15 – 17:45 Lightning talks for poster session 1 (odd numbers) 17:45 – 19:15 Poster session 1 (odd numbers) with drinks reception 19:15 – 21:00 Dinner – buffet Thursday, 24 May 2018 08:30 – 10:30 Session 3: Variant interpretation 10:30 – 11:00 Morning coffee 11:00 – 12:30 Session 4: Somatic variation 12:30 – 14:00 Lunch and meet the speakers 14:00 – 15:45 Session 5: Next generation phenotyping 15:45 – 16:15 Afternoon tea 16:15 – 17:45 Session 6: Gene curation 17:45 – 18:15 Lightning talks for poster session 2 (even numbers) 18:15 – 19:45 Poster session 2 (even numbers) with drinks reception 19:45 – 22:00 Conference dinner – served Friday, 25 May 2018 08:30 – 10:30 Session 7: Considerations for population testing 10:30 – 11:00 Morning coffee 11:00 – 12:45 Session 8: Reanalysis 12:45 – 13:00 Closing remarks 13:00 – 13:15 Take away lunch 13:15 Coaches depart to Cambridge city centre via train station, and
Heathrow airport via Stansted airport
8
Conference Sponsor We would like to acknowledge the generous support from the following organisations:
www.congenica.com
www.genomediagnosticsnijmegen.nl
9
Curating the Clinical Genome 2018
Wellcome Genome Campus Conference Centre
Hinxton, Cambridge
23 – 25 May 2018
Lectures to be held in the Francis Crick Auditorium Lunch and dinner to be held in the Hall Restaurant
Poster sessions to be held in the Conference Centre
Spoken presentations - If you are an invited speaker, or your abstract has been selected for a spoken presentation, please give an electronic version of your talk to the AV technician.
Poster presentations – If your abstract has been selected for a poster, please display this in the Conference Centre on arrival.
Conference Programme
Wednesday 23 May 2018 12:00-13:00 Registration with lunch 13:00-13:15 Welcome and introduction
Helen Firth Cambridge University Hospitals, UK 13:15-15:00 Session 1: Data Sharing
Chair: Heidi Rehm, Harvard Medical School, USA
13:15 The future of health and research data in genomics Ewan Birney
EMBL-EBI, UK 13:45 The genomic glass house: Data sharing, individual data access,
and civil rights Barbara Evans
University of Houston, USA 14:15 Our Genematcher data sharing experience: 10 days on average
to confirm the pathogenicity of a candidate gene Ange-Line Bruel INSERM U1231, France
14:30 DECIPHER – Innovation in data-sharing in rare disease
Julia Foreman Wellcome Sanger Institute, UK
14:45 Discussion
10
15.00-15:30 Afternoon tea 15:30-17:15 Session 2: Variant Guidelines and Resources
Chair: Dominic McMullan, West Midlands Regional Genetics Service, UK 15:30 A systematic framework for the interpretation of copy number
variants Christa Martin
Geisinger, USA 16:00 Assessing the global landscape of clinical genetic variation Gillian Belbin
Mount Sinai, USA 16:30 Improving Ensembl’s resources for genomic interpretation
Fiona Cunningham EMBL-EBI, UK
16:45 UniProtKB/Swiss-Prot in the era of personalized medicine:
Current work on variant interpretation and annotation Maria Livia Famiglietti SIB Swiss Institute of Bioinformatics, Switzerland
17:00 Discussion
17:15-17:45 Lightning talks Chair: Marc Williams, Geisinger, USA
Thursday 24 May 2018 08:30-10:30 Session 3: Variant Interpretation Chair: Christa Martin, Geisinger, USA
08:30 Disease-specific optimisation of variant interpretation Nicola Whiffin
Imperial College London, UK 09:00 Common and rare genetic variants and the risk of breast cancer Antonis Antoniou
University of Cambridge, UK 09:30 The NIHR BioResource experience: Variant interpretation in
10,000 Whole Genome Sequenced DNA samples Karyn Megy University of Cambridge, UK
11
09:45 The ClinGen Storage Disorders Expert Panel’s guidelines for GAA variant interpretation: Towards improved Pompe disease diagnostics Jennifer Goldstein UNC / ClinGen, USA
11:00 Interpreting the cancer genome Serena Nik-Zainal
University of Cambridge, UK 11:30 Cancer genome interpreter annotates the biological and clinical
relevance of tumor alterations David Tremborero
UPF / IRB / Karolinska, Spain
12:00 COSMIC, an essential resource for the clinical interpretation of cancer genomes
Ray Stefancsik Wellcome Sanger Institute, UK 12:15 Discussion
12:30-14:00 Lunch and meet the speakers Hall Restaurant 14:00-15.45 Session 5: Next Generation Phenotyping Chair: Helen Firth, Cambridge University Hospitals, UK
14:00 Assessing specificity in phenotypic spectra associated with
molecularly-defined human developmental disorders David FitzPatrick
University of Edinburgh, UK 14:30 Electronic health record phenotyping: An emerging science Peggy Peissig
Marshfield Clinic Research Institute, USA
15:00 Defining and refining disease nomenclature based on gene-focused curations in the age of genomic medicine Courtney Thaxton ClinGen / UNC, USA
12
15:15 Exome sequencing of 506 parental/fetal trios with structural
abnormalities revealed by ultrasound in the UK Prenatal Assessment of Genomes and Exomes (PAGE) project
Dominic McMullan West Midlands Regional Genetics Service, UK 15:30 Discussion
Chair: David FitzPatrick, University of Edinburgh, UK 16:15 Reappraisal of reported genes for sudden arrhythmic death: An
evidence-based evaluation of gene validity for Brugada syndrome Michael Gollob
University of Toronto, Canada
16:45 Curating clinically relevant transcripts for the interpretation of sequence variants
Marina DiStefano Partners Healthcare Personalized Medicine, USA 17:00 Implementation of gene curation in a clinical laboratory setting Alison Coffey Illumina, USA 17:15 Assessing the strength of evidence for genes implicated in fatty
acid oxidation disorders using the ClinGen Clinical Validity Framework
Jennifer McGlaughon UNC / ClinGen, USA 17:30 Discussion
17:45-18:15 Lightning talks
Chair: Marc Williams, Geisinger, USA 18:15-19:45 Poster session 2 (even numbers) with drinks reception
19:45 Dinner
Hall Restaurant
Friday 25 May 2018 08:30-10:30 Session 7: Considerations for Population Testing
Chair: Gert Matthijs, KU Leuven, Belgium
08:30 Balancing the sensitivity and specificity of variant classification for healthy populations
Peter Kang Counsyl, USA
13
09:00 Genetic cascade screening for Familial Hypercholesterolemia: a national cardiovascular disease prevention programme
Joep Defesche Academic Medical Centre, University of Amsterdam, The Netherlands
09:30 Clinical interventions to delay or prevent outcomes related to
inherited conditions: Do expert opinions on the nature of intervention reflect the opinions of the general population?
Katrina Goddard Kaiser Permanente, USA 09:45 Panel session/open discussion
10:30-11:00 Morning coffee 11:00-12:45 Session 8: Reanalysis Chair: Caroline Wright, University of Exeter, UK
11:00 Clinical whole-exome sequencing for the diagnosis of rare disorders with congenital anomalies and/or intellectual disability: substantial interest of prospective annual reanalysis
Sophie Nambot University of Dijon, France
11:30 Implementation of a whitelisting approach to make additional
diagnoses of single-gene developmental disorders in whole exome trios Panayiotis Constantinou
Addenbrooke's Hospital, UK 11:45 Scaling the resolution of sequence variant interpretation
discrepancies in ClinVar Steven Harrison Harvard Medical School, USA
12:00 GenomeConnect: Sharing individual level data through patient
registries Juliann Savatt
Geisinger, USA 12:30 Discussion
12:45-13:00 Closing remarks Helen Firth Cambridge University Hospitals, UK
Heidi Rehm Harvard Medical School, USA
13:00-13:15 Take away lunch 13:15 Coaches depart to Cambridge city centre (Downing Street) via train
station, and Heathrow airport via Stansted airport
Sapientia by Congenica is a clinical decision support software for analysis, interpretation and generation of clinically actionable reports on patient derived genomic data.
With underlying technology that spun out of the pioneering research from the Sanger Institute, our platform provides clinical genome analytics to support medical practitioners treating patients with genetic diseases, and thereby improving human health and personalised patient care.
These abstracts should not be cited in bibliographies. Materials contained herein should be treated as personal communication and should be cited as such only with
consent of the author.
16
Notes
S1
Spoken Presentations
The future of health and research data in genomics
Ewan Birney
EMBL-EBI, UK
Molecular biology is now a leading example of a data intensive science, with both pragmatic
and theoretical challenges being raised by data volumes and dimensionality of the data.
These changes are present in both “large scale” consortia science and small scale science,
and across now a broad range of applications – from human health, through to agriculture
and ecosystems. All of molecular life science is feeling this effect.
As molecular techniques – from genomics through transcriptomics and metabolomics – drop
in price and turn around time there is a wealth of opportunity for clinical research and in
some cases, active changes clinical practice even at this early stage. The development of
this work requires inter-disciplinary teams spanning basic research, bioinformatics and
clinical expertise.
This shift in modality is creating a wealth of new opportunities and has some accompanying
challenges. In particular there is a continued need for a robust information infrastructure for
molecular biology and clinical research. This ranges from the physical aspects of dealing
with data volume through to the more statistically challenging aspects of interpreting it.
A particular opportunity is the switch from research commissioning genomic measurement to
healthcare centric genomic measurement. This is occurring in a number of countries
worldwide, including Australia, Denmark, Finland, France, United Kingdom and United
States. The Global Alliance for Genomics and Health provides a standards setting
organisation to allow for both a deepening of the technical aspects of healthcare and
allowing for appropriate secondary use for research of healthcare commissioned genomics
data.
I will outline the overall challenge present in this new, interdisciplinary field and the global
coordination needed to achieve its goals.
S2
Notes
S3
The genomic glass house: Data sharing, individual data access, and civil rights
Barbara Evans
University of Houston, USA
Genomic testing is a magnificent and brightly sparkling glass house, an edifice built of data
transparency. It will require transparent and, at times, unconsented sharing of people’s data
to create inclusive data commons that support vibrant scientific discovery, judicious and
trustworthy regulatory oversight, and well-informed clinical translation of genomic test
results. Transparency confers many social benefits but threatens the civil rights of the
hapless souls whose genomic information will be shared. This presentation retraces the 900-
year-long history of large-scale data commons, dating back to the collection of data for
England’s Domesday Book in 1085-1086. History confirms that the modern norm of informed
consent for clinical research reflects longstanding legal and cultural traditions against
unconsented touching of the human body. Yet if there are any longstanding legal and
historical norms relating to data, they seem to favor unconsented collection and use. The
notion that people should be asked to consent to socially beneficial uses of their data
appears to spring forth late in the 1970s, as Baby Boomers came of age and struggled—
and, by some accounts, we failed—to draw a sensible line between autonomy and
narcissism. Lawyers, judges, and legislators kept a cool head, however, and the legal
frameworks of all major jurisdictions continue to allow access to data for important research,
public health, and regulatory purposes. This access raises a question, though: What ethical
duties do we owe to people whose personal information is used, without their consent, to
build genomic data commons for research and other socially beneficial purposes? Many
people, if asked, might simply reply that their sensitive data should not be used without
consent, but that is not the question. The question is how to make the process of
unconsented data access as ethical and as worthy of public trust as it possibly can be.
Designing alternative protections—a set of ethical standards for unconsented data use, as it
were—feels like an oxymoron or a second-best solution, akin to debating the most ethical
way to poison a baby. We all feel reluctant to deliberate ethical alternatives to consent, lest
doing so suggest our complicity in undermining the hallowed—if largely imaginary—informed
consent norms that people love but law does not requite. This talk revisits past attempts to
define ethical standards for unconsented use of people’s data. Several principles emerge.
One major principle is that data transparency implies sharing data not just with third parties
such as researchers, regulators, public health officials, and clinicians who want to use
people’s data, but with the people whose data are being shared. If you want access to
people’s data, you must grant them access, too. Take it or leave it; that is the only ethical
deal that is available.
S4
Notes
S5
Our Genematcher data sharing experience: 10 days on average to confirm the
pathogenicity of a candidate gene
Ange-Line Bruel1,2, Antonio Vitobello1,2, Fred Tran Mau-Them1,2, Sophie Nambot1,2,3,
Virginie Quéré1,2, Paul Kuentz1, Julien Thevenon1,3, Mirna Assoum1, Sébastien
Laure Mosca-Boidron1,2, Patrick Callier1,2, Christophe Philippe1,2, Laurence Faivre1,3,
Christel Thauvin-Robinet1,2,3,4
1- UMR1231 GAD, Inserm - University of Burgundy-Franche Comté, Dijon, France
2- Unit for Innovation and Genomic Diagnosis of Rare Diseases, FHU-TRANSLAD, Dijon
University Hospital, Dijon, France
3- Centre of Reference for Rare Diseases: Development disorders and malformation
syndromes, Genetics Department, FHU-TRANSLAD, Dijon University Hospital, Dijon,
France
4- Centre of Reference for Rare Diseases: Development disorders, Genetics Department,
FHU-TRANSLAD, Dijon University Hospital, Dijon, France
Whole-exome sequencing (WES) has proven to be a powerful tool to identify the molecular
bases of heterogeneous conditions such as intellectual disability (ID) and/or multiple
congenital abnormalities (MCA). A large number of results remain non-conclusive, especially
for ultra-rare conditions that limit genotype-phenotype correlations. International data-sharing
was used to identify additional patients carrying variants in the same gene, in order to draw
definitive conclusions on their implication in the disease. Here, we report our experience
using the GeneMatcher initiative, a data-sharing platform designed to enable connections
between clinicians and researchers to help solve 'unsolved' exomes and to identify new
genes. Over the last two years, we have shared 71 candidate genes identified by WES
performed in individuals affected by ID/MCA. We evaluated the ability to determine the
involvement of these genes and the necessary timeframe: 60/70 genes (85%) were matched
to at least one other mutated individual, and 24 genes recurring in additional affected
individuals were identified as the probable cause of a developmental disease (39%). The
waiting period between submission and the first match varied, with an overall median of 4
hours. When a match occurred, the median response time between the first email to contact
a submitter and the response was estimated at 31 hours. The rapid identification of these
new genes remains essential for clinical characterization, genetic counselling and for
translation to the diagnostic field. GeneMatcher appears to be a very efficient tool to identify
new genes in highly heterogeneous conditions.
S6
Notes
S7
DECIPHER – Innovation in data-sharing in rare disease
Julia Foreman, A Paul Bevan1, Simon Brent1, Ben Hutton1, Daniel Perrett1, Kaitlin Samocha1, Matthew E Hurles2 and Helen V Firth1,2
1 Wellcome Sanger Institute, Wellcome Genome Campus, Cambridge, Hinxton CB10 1SA 2 Cambridge University Hospitals NHS Foundation Trust, Cambridge Biomedical Campus, Hills Road, Cambridge CB2 0QQ
DECIPHER (https://decipher.sanger.ac.uk) established in 2004 has grown to become a
major global platform for the visualization of phenotypic and genomic relationships and for
sharing rare disease patient records. DECIPHER displays all scales of genomic variation,
from single base to megabases, in a single interface. DECIPHER's mission is to map the
clinically relevant elements of the genome and understand their contribution to human
development and disease. The website provides users with interfaces to assist in the
interpretation of variant pathogenicity, and includes a genome browser, a protein browser
and matching patient interface. In addition DECIPHER offers an ACMG pathogenicity
evidence interface which allows users to record and share the evidence for sequence variant
classification. Phenotypic information is also aggregated to inform about the natural history
of disease associated with pathogenic variants in clinically relevant genes.
DECIPHER has established a global network of participating projects and now has >240
projects in 6 continents. There are over 25,500 patient records in DECIPHER with open-
access consent, searchable via the openly available search engine on the homepage by
phenotype terms, gene name and genomic coordinates. DECIPHER promotes flexible data-
sharing, enabling the extent of sharing to be tailored so that it is proportionate to the clinical
or scientific need to facilitate diagnosis or discovery.
DECIPHER is a pioneering partner in the Global Alliance for Global Health (GA4GH) and
has driven the development of the Matchmaker exchange Application Programming
Interface (API) enabling the federated discovery of similar patients in connected databases
for a given patient in DECIPHER (currently Phenome Central, Broad MatchBox, MyGene2
and GeneMatcher).
S8
Notes
S9
A systematic framework for the interpretation of copy number variants
Christa Lese Martin
Geisinger, USA
Analysis of germline copy number variants (CNVs), including deletions and duplications, is
an established first-tier evaluation in individuals with neurodevelopmental disorders and/or
multiple congenital anomalies. Although whole-genome CNV analysis has been in routine
diagnostic use for almost a decade, interpretation of the clinical significance of some CNVs
remains challenging. The clinical interpretation of CNVs has also become increasingly
sophisticated in recent years due to the discovery of novel genomic disorders, studies
providing a deeper view of the broad phenotypic spectrum in individuals with CNVs, and
innovations in high-resolution microarray and sequencing technologies that allow the
detection of CNVs with resolution spanning a single gene to entire chromosomes. The
process of CNV interpretation relies on the meaningful use of appropriate evidence to
support or refute pathogenicity. Despite progress in making CNV interpretation consistent in
recent years, discordance persists among clinical laboratories, and is largely attributable to
differences in selecting and weighing the evidence used in classifying clinical significance.
While existing ACMG guidelines provide a high-level conceptual framework for applying
evidence to interpretations of constitutional CNVs in diagnostic testing, more refined
specifications are needed to promote consistency and transparency in CNV interpretation.
The ACMG and the Clinical Genome Resource (ClinGen) are collaborating to update these
guidelines with more specific recommendations on how and when to account for various
types of evidence. We have devised a point-based, hierarchical scoring system to
systematically evaluate relevant evidence to determine the pathogenicity of CNVs, including
overlap with CNVs reported in clinically affected individuals, overlap with CNVs reported in
unaffected individuals, case-control studies, the presence of known dosage-sensitive genes,
case reports with segregation data, de novo occurrence of CNVs, and the number of protein-
coding genes included in the CNV. We have developed a detailed rubric for interpreting
deletions and duplications and are testing this framework with a broad group of
cytogeneticists to identity nuances and fine-tune its guidance. The updated guidelines will
apply to analysis performed with any method capable of defining the chromosomal
boundaries of copy number events, even routine whole-exome or genome sequencing. This
work is expected to have broad impact in the clinical community by providing a robust
system to support the consistent interpretation of CNVs.
S10
Notes
S11
Assessing the global landscape of clinical genetic variation
Gillian Belbin
Mount Sinai, USA
Characterization of Mendelian disease has benefitted from emerging large, open-source
genomic databases. We highlight the value of adding increased global diversity and fine-
scale population substructure to databases used for annotating clinical variants (CVs). As
part of the Population Architecture using Genomics and Epidemiology (PAGE) Study, we
genotyped 63,131 CVs from clinical databases in 51,698 individuals, comprising 99 global
populations. We showed that increasing population diversity, and addressing systematic
biases in medical databases, enabled the filtering of 40% more variants using medical
guidelines. Additionally, we linked genetic disease loci to health records for 60 populations in
the BioMe Biobank in New York City and detected a common (1:23) hypertrophic
cardiomyopathy variant in a founder population from Central America. Using PheWAS
approaches, we also defined a broad range (1:40-1:8) of incidental genetic disease in
BioMe. This work has ramifications for medical genetics and clinical care, improving
precision medicine for the world.
S12
Notes
S13
Improving Ensembl’s resources for genomic interpretation
Fiona Cunningham, Irina Armean1, Alexander Astashyn2, Ruth Bennett1, Claire Davidson1, Helen V Firth3, David R FitzPatrick4, Adam Frankish1, Laurent Gil1, Mihail Halachev4, Sarah Hunt1, Vinita Joardar2, Mike Kay1, Jane Loveland1, Kelly McGarvey2, William McLaren1, Aoife McMahon1, Joannella Morales1, Terence Murphy2, Andrew Parton1, Helen Schuilenburg1, Anja Thormann1, Glen Threadgold1, Caroline F Wright5.
1 European Molecular Biology Laboratory, European Bioinformatics Institute, Hinxton, UK; 2 National Center for Biotechnology Information, U.S. National Library of Medicine 8600 Rockville Pike, Bethesda MD, USA; 3 Cambridge University Hospitals, Cambridge, UK; 4 MRC Human Genetics Unit, IGMM, Edinburgh, UK; 5 University of Exeter Medical School, RILD Level 4, Royal Devon & Exeter Hospital, Barrack Road, Exeter, UK.
Ensembl's Variant Effect Predictor (VEP) is a tool for variant annotation, interpretation and
analysis. The VEP integrates variant, phenotypic, regulatory and transcript data from
Ensembl to annotate individual genomes or specific variants. The software is customisable
and can be configured to use private datasets. The VEP can annotate a human genome (of
around 4M variants) in 30 mins and an exome in only 10 mins.
Improving transcript representation is critical for the interpretation of genomic data. Recently,
we changed the VEP's handling of RefSeq transcripts to correctly annotate variants when
the RefSeq transcript does not match the reference genome. We continue to work with the
NCBI to cross-compare annotation. We aim to select one identical reference transcript in
both the Ensembl/GENCODE and RefSeq transcript set, and work with the clinical
community to have this represented as an Locus Reference Genomic (LRG) sequence.
To facilitate targeted filtering of annotated variants from VEP, we developed a VEP
extension, a 'plugin' that utilises gene-disease panel information. Developed in collaboration
with the Deciphering Developmental Disorders (DDD) project, the VEP-G2P plugin works
with our gene-disease panel website, gene2phenotype* but can use comparable data from
any gene panel. The plugin filters VEP results to only return variants that: 1) match the allelic
requirement specified for the gene; 2) pass a customisable allele frequency threshold; and 3)
have a severe mutation consequence. Although agnostic of inheritance or patient
phenotypes, using the DDD patient set the VEP-G2P plugin selected over 80% of known
causative de novo mutations. The false positive rate was 23% indicating a careful review of
UniProtKB/Swiss-Prot in the era of personalized medicine: Current work on variant interpretation and annotation
M.L. Famiglietti [1], A. Estreicher [1], L. Breuza [1], S. Poux [1], N. Redaschi [1], I. Xenarios [1,2,3], A. Bridge [1], and the UniProt Consortium [1,4,5]
[1] Swiss-Prot group, SIB Swiss Institute of Bioinformatics, CMU, 1 rue Michel Servet, 1211 Geneva 4, Switzerland. [2] Vital-IT Group, SIB Swiss Institute of Bioinformatics, Quartier Sorge, Bâtiment Génopode, 1015 Lausanne, Switzerland. [3] University of Lausanne, 1015 Lausanne, Switzerland. [4] European Bioinformatics Institute (EBI), Hinxton, UK. [5] Protein Information Resource (PIR), Washington, DC, USA.
In the era of personalized genomic medicine advances in human healthcare will be powered
by the integration of many data sources and types, including an individual's genomic and
phenotypic data and reference data on the functional and clinical significance of
experimentally characterized DNA sequence variants from the wider population.
UniProtKB/Swiss-Prot is one resource of genetic variant information, providing over 78,000
expert curated missense variants extracted from the literature, including over 30,000
mutations implicated in mostly Mendelian diseases. Here, we present a pilot study on the
alignment of UniProtKB/Swiss-Prot variant interpretations with those provided by ClinVar
and ClinGen. ClinVar and ClinGen employ a 5-tier classification of variant significance, as
proposed by the American College of Medical Genetics and Genomics and the Association
for Molecular Pathology (ACMG-AMP): pathogenic, likely pathogenic, benign, likely benign,
uncertain significance, while UniProt employs a simpler classification scheme, which
basically equates to pathogenic, benign, or uncertain significance. We limited our initial
comparisons to ClinVar 2-star-records, i.e. records with concordant interpretations from
multiple submitters but not yet reviewed by an expert panel. A preliminary comparison of
some 2,000 UniProtKB/Swiss-Prot variant interpretations with those of the corresponding
ClinVar 2-star records indicated that around 10% of variant interpretations in
UniProtKB/Swiss-Prot differed from those in ClinVar. When the ClinGen Pathogenicity
Calculator was used to apply ACMG/AMP criteria to the UniProt workflow, 70% of these
discrepancies could be resolved. These preliminary results highlight the need for community
guidelines, robust tools, and continuous re-evaluation of clinical variant data as datasets and
knowledge improve.
S16
Notes
S17
Disease-specific optimisation of variant interpretation
Nicola Whiffin
Imperial College London, UK
Incorporation of gene- and disease-specific data is important for accurate clinical variant
interpretation. Although determining evidence-based thresholds for activating criteria
outlined in the 2015 ACMG/AMP guidelines can be difficult for very rare disease, emerging
guidance based on more common phenotypes presents the opportunity to learn from
disorders with similar genetic architectures.
Inherited heart conditions (ICCs) are a group of dominant disorders characterised by adult
onset and reduced penetrance. At a combined population prevalence of ~1%, ICCs are
common enough to permit empirical evaluation of ACMG/AMP thresholds.
Using large cohorts of diseased individuals and both population and healthy controls, we
have refined the ACMG/AMP guidelines for use in cardiovascular phenotypes. One such
refinement is the development of disease-specific frequency thresholds. Our rigorous
statistical framework, which considers disease prevalence, genetic heterogeneity and variant
penetrance, evaluates whether an observed allele frequency is compatible with
pathogenicity. In addition to permitting the safe and appropriate use of much more stringent
AF thresholds, our approach facilitates investigation of disease architecture, including
accurate estimation of variant penetrance.
In addition, we have created wed-based tools to aid visualisation and comparison of these
datasets, and to improve the accuracy and reproducibility of cardiac variant interpretation.
These include CardioClassifier (cardioclassifier.org), an automated and interactive web-tool
to aid interpretation of variants in genes associated with inherited cardiac diseases.
S18
Notes
S19
Common and rare genetic variants and the risk of breast cancer
Antonis Antoniou
University of Cambridge, U.K.
Advances in genomic technologies have enabled more rapid, less expensive genetic
sequencing than was possible a few years ago. These technologies allow for the
comprehensive genetic profiling for assessing risks to breast cancer and include multiplex
sequencing panels of several genes and panels of common single nucleotide
polymorphisms (SNPs). However, the clinical utility of such multiplex gene and SNP panels
depends on having accurate estimates of cancer risks for mutations in the genes included in
such panels as well as cancer risk prediction models that consider the multifactorial
aetiology to cancer susceptibility. Over the past decade international consortia, such as the
Breast and Cancer Association Consortium, the Consortium of Investigators of Modifiers of
BRCA1/2, the International BRCA1/2 Carrier Cohort Study and the PALB2 Interest Group
have enabled us to accurately characterise the cancer risks for rare and common cancer
susceptibility genetic variants; to understand how the genetic variants interact with each
other; and how genetic variants interact with other lifestyle/hormonal risk factors for the
disease. The presentation will review the key recent advances by these international
consortia and how these are helping us to realise a more personalised risk-based cancer
prevention and cancer control.
S20
Notes
S21
The NIHR BioResource experience: Variant interpretation in 10,000 Whole Genome Sequenced DNA samples
Karyn Mégy(1), Rutendo Mapeta(1,2), Sri VV Deevi(1,2), Christopher J Penkett(1,2), Kathleen Stirrups(1,2), Lucy F Raymond(1,3,4), Willem H Ouwehand (1,2,4,5), NIHR BioResource(1)
(1)NIHR BioResource, University of Cambridge, Cambridge, UK. (2)Department of Haematology, University of Cambridge, Cambridge, UK. (3)Department of Medical Genetics, Cambridge Institute for Medical Research, Cambridge, University of Cambridge, UK. (4)Wellcome Sanger Institute, Hinxton, Cambridge, UK. (5)NHS Blood and Transplant, Cambridge Biomedical Campus, Cambridge, UK.
The analysis of 10,000 DNA samples by Whole Genome Sequencing (WGS) from patients and their close relatives has been completed as part of the pilot phase for rare diseases of the Genomics England 100 000 Genomes Project. The aims are to identify the genetic causes, improve rates of molecular diagnosis and enable research, identify novel associated genes and develop new treatments for 15 rare disease categories. A first pass pertinent finding analysis has been completed on all samples. The interpretation of variants followed the ACMG guidelines to ensure consistency between disease categories. Experts in each of the disorders produced a curated list of pertinent (known) genes for submission to the Locus Reference Genome (LRG) initiative in order to select a representative transcript(s). Variants in the pertinent genes were prioritised based on frequency (gnomAD MAF <0.001 if novel; gnomAD MAF <0.025 if in HGMDPro), consequence (splice region; non-coding exon if in a non-coding RNA gene; high or moderate impact according to VEP) and the resulting list was passed to a Multi-Disciplinary Team (MDT), composed of a clinical consultant and experts in clinical genetics and bioinformatics, who assessed the prioritised variants in the context of the human phenotype ontology (HPO) encoded clinical and laboratory phenotype data. While the programmatic prioritisation efficiently reduces the number of variants per patient from thousands to less than 10, the critical decision by the MDT whether a variant (V) is pathogenic or likely pathogenic one (PV or LPV respectively) and explains the patient's phenotype remains a process requiring expert MDT opinion. The diagnostic yield ranges between the disorder categories from 1% to 55%, reflecting the type of disorders and particularly the eligibility criteria applied for enrolment (e.g. for certain categories of diseases cases were enrolled if no PVs or LPVs could be identified in known relevant genes). Interestingly, and consistently across diseases, over half of the 1,581 PVs/LPVs identified were novel. So far, a conclusive molecular diagnosis was generated for nearly 1,300 patients and about 25 new genes have been discovered, demonstrating the feasibility of WGS analysis in the clinical setting, which, together with HPO terms lead to a doubling of the number of causal variants in known genes and substantially increased the number of genes implicated in rare diseases.
S22
Notes
S23
The ClinGen Storage Disorders Expert Panel’s guidelines for GAA variant interpretation: Towards improved Pompe disease diagnostics
Jennifer Goldstein1, Catherine Rehder2, Steven Harrison3, Heather Baudet1, Jixia Liu4, Miriam Schachter5, Jennifer McGlaughon1, Bryce Seifert1, Yue Si6, Courtney Thaxton1, Rupa Udani7, Meredith Weaver8, Deeksha Bali2, Michele Caggana9, Madhuri Hegde10, Michael Watson8, Robert Steiner7
1UNC, Chapel Hill, NC, USA; 2Duke University Health System, Durham, NC, USA; 3Partner’s Healthcare, Cambridge, MA, USA; 4Marshfield Clinic, Marshfield, WI, USA; 5New Jersey Department of Health, Ewing, NJ, USA; 6GeneDx, Gaithersburg, MD, USA; 7University of Wisconsin, Madison, WI, USA; 8ACMG, Bethesda, MD, USA; 9New York State Health Department, Albany, NY, USA; 10PerkinElmer, Inc., USA.
Publication of the ACMG-AMP criteria in 2015 was an important step towards standardizing
variant interpretation. However, as these criteria were designed to apply to a wide range of
Mendelian disorders, gene- and disease-specific specifications are needed to limit the
potential for interpretive discrepancies to occur. To address this, the Clinical Genome
Resource (ClinGen) is assembling expert panels to adapt the ACMG-AMP criteria for
interpretation of variants in genes of interest, and to submit the interpretations to the publicly
available ClinVar database. The ClinGen Storage Disorders Expert Panel has specified the
ACMG-AMP criteria for interpretation of variants in GAA, the gene associated with Pompe
disease (glycogen storage disease type II; acid maltase deficiency). Newborn screening for
Pompe disease has been approved by the Secretary's Advisory Committee in the USA, and
several countries are now screening for this condition at birth. Accurate interpretation of
variants within GAA is important for confirmation of the diagnosis of patients of any age,
including asymptomatic infants identified by newborn screening, and for diagnostic and
carrier testing for family members. We will describe our rules for GAA variant interpretation.
This includes specifications to ACMG-AMP criteria based on the characteristics of GAA and
Pompe disease, criteria that our group deemed applicable to be used "as is", and those
criteria considered to be inapplicable to GAA variant interpretation. We will describe an initial
pilot study of 16 variants to test the clarity and validity of our draft GAA variant interpretation
guidelines. Our long-term goal is to complete a larger pilot study to fully validate the
guidelines, to use these criteria for interpretation of reported GAA variants, and to submit
those interpretations to ClinVar for use by the scientific and medical communities.
S24
Notes
S25
ClinGen cardiomyopathy expert panel, phase 2: Implementation of sustained variant curation and classification C. Lisa Kurtz1, Melissa A. Kelly2; Christina Austin-Tse3; Kate Thomson4 and Birgit Funke5 1UNC Chapel Hill, Chapel Hill, NC, USA; 2Geisinger, Danville, PA, USA; 3Laboratory for Molecular Medicine, Cambridge, MA, USA; 4Oxford University Hospital, Oxford, UK; 5Veritas Genetics, Boston, MA, USA
As the first approved ClinGen internal EP, the ClinGen cardiomyopathy expert panel (CMP-
EP) specified the ACMG/AMP variant classification guidelines for MYH7 and is approved to
submit classifications to ClinVar at the 3-star status level with the goal of using these
adjusted rules to classify all hypertrophic cardiomyopathy (HCM)-associated MYH7 variants
in the public domain. The membership continues to cover clinical and molecular testing
expertise and now includes 28 members representing four countries.
In collaboration with the UK's Association for Clinical Genomic Science (ACGS), the CMP-
EP has outlined a process to tackle this monumental project and recruited additional variant
experts to serve the needs of a sustained variant curation effort, increasing the number of
represented academic and commercial cardiovascular testing laboratories to 10. Two
experts from each laboratory (typically one senior and one junior member) were invited to
join. Participation requirements included the ability to join regular monthly phone calls and,
for laboratory members, a commitment to curate a minimum number of variants each month.
The CMP-EP continues to play a strategic role, specifying guidelines for HCM-associated
genes beyond MYH7 and adjudicating variant classifications as needed, while an
independent but overlapping variant curation committee (VCC) will carry out variant curation.
A priority list of 581 variants was pulled from the >1,700 MYH7 variants present in ClinVar,
HGMD and LOVD, by the core team, which also developed a curation process. Each variant
will undergo dual review with labs initially using their own internal variant classification
processes to gather evidence and then applying the adapted MYH7 rules. Members of the
core team will review curations and route any issues that require discussion (such as
discordant classifications) to the CMP-EP for final approval, followed by submission to
ClinVar at the 3-star status. When MYH7 variant classification is complete, the CMP-
EP/VCC will apply the same process to all HCM-associated genes/variants using newly
specified guidelines.
As the first approved internal ClinGen EP, the CMP-EP is pioneering the architecture and
logistics of sustained variant curation, but adequate funding and manpower remain critical
challenges to enabling completion of this work in a meaningful timeline.
S26
Notes
S27
Interpreting the cancer genome
Serena Nik-Zainal
University of Cambridge, UK
Abstract not available at the time of printing
S28
Notes
S29
Cancer genome interpreter annotates the biological and clinical relevance of tumor
alterations
David Tamborero
UPF/IRB/Karolinska, Spain
While tumor genome sequencing has become widely available in clinical and research
settings, the interpretation of tumor somatic variants remains an important bottleneck. Here
we present the Cancer Genome Interpreter, a versatile platform that automates the
interpretation of newly sequenced cancer genomes, annotating the potential of alterations
detected in tumors to act as drivers and their possible effect on treatment response. The
results are organized in different levels of evidence according to current knowledge, which
we envision can support a broad range of oncology use cases. The resource is publicly
available at http://www.cancergenomeinterpreter.org. Of note, the database of genomic
biomarkers of drug response is under continuous update by a board of medical oncologists
and cancer genomics experts. This effort is currently integrated with the projects of other
leading institutions developing these knowledge bases by the Variant Interpretation for
Cancer Consortium (http://cancervariants.org/) under the umbrella of the Global Alliance for
Genomics & Health. Besides the aggregation of the data collected by each individual
resource, the aim of this project will be to establish community standards to represent and
share this information.
S30
Notes
S31
COSMIC, an essential resource for the clinical interpretation of cancer genomes
Ray Stefancsik, Dave Beare, Nidhi Bindal, John Tate, Christopher Ramshaw, Charlie Hathaway, Charalampos Boutselakis, Chai Yin Kok, Shicai Wang, Bhavana Harsha, Sally Bamford, Charlotte Dunham, Elisabeth Dawson, Sari Ward, Steven Jupe, Laura Ponting, Harry Jubb, Samantha Thompson, Zbyslaw Sondka, Kate Noble, Claire Rye, Simon A. Forbes
Wellcome Sanger Institute, UK
COSMIC, the Catalogue Of Somatic Mutations In Cancer, is a vast collection of expert
curated somatic mutations, associated clinical features, environmental risk factors and other
cancer-relevant information that makes it the most comprehensive resource of its kind.
There are several ways of how COSMIC can help the interpretation of somatic variants from
clinical cancer samples.
Somatic variants included in COSMIC from cancer samples are curated from biomedical
literature and large scale multi-regional studies of cancer genomes by expert curators. The
Cancer Gene Census (CGC) is a regularly updated catalogue of those genes which contain
mutations that have been causally implicated in cancer. Once the literature for a census
gene has been completely curated, it is released and included in the list of 'COSMIC classic'
genes. Recurrence frequencies of gene mutations and the associated tumour types together
with the CGC makes COSMIC a great resource for creating multi-gene panels for screening
and diagnostic testing of clinical samples from various cancer types. The usefulness of
COSMIC in clinical cancer genome interpretation is attested by the fact that several
diagnostic providers already utilise COSMIC cancer variant data in their workflow.
COSMIC-3D can help make inferences about the structural and functional consequences of
particular mutations in proteins thereby facilitating drug target identification.
Some somatic mutations allow tumour cells to evade therapeutic cancer drugs. COSMIC
provides drug resistance information to facilitate diagnostic and pharmaceutical research
and development in this area. The frequency of mutations in the evolution of therapeutic
drug resistance, and additional mutational and clinical information associated with resistant
samples can be explored in COSMIC.
Additionally, useful information is available from COSMIC on non-coding variants, abnormal
copy number segments and epigenetic changes at CpG dinucleotides in cancer genomes.
All the accumulated knowledge in COSMIC is available freely on its website. Moreover, the
'Data Downloads' section allows bulk access to the various COSMIC data files for registered
users making it altogether the most comprehensive web-resource for exploring the clinical
impact of somatic mutations in human cancer.
S32
Notes
S33
Assessing specificity in phenotypic spectra associated with molecularly-defined
human developmental disorders
David R FitzPatrick1, Stuart Aitken1, Caroline Wright2,3, Matt Hurles2, Helen Firth2,4
1.MRC Human Genetics Unit, University of Edinburgh 2.Wellcome Sanger Insititute, Hinxton,
Cambridge. 3.Institute of Biomedical and Clinical Science, University of Exeter Medical
School 4.Clinical Genetics Department, Addenbrookes Hospital, Cambridge
Clinically-defined syndrome diagnoses have an excellent record in predicting defined sets of
causative genotypes. The Deciphering Developmental Disorders (DDD)1,2 project is a UK-
and Ireland-wide study that aims to develop and use new genetic technology and statistical
analyses to make a definitive diagnosis in individuals with severe or extreme developmental
disorders. DNA samples are available from ~13,500 affected individuals have been
recruited with 10,000 of these also having samples available from both parents. Using
human genetic data alone DDD has established that damaging de novo variants in
monoallelic developmental disorder genes are the major cause of previously undiagnosed
developmental disorders2. Such variants have a positive predictive diagnostic value of~0.83.
The scale and diversity of the DDD cohort together with the systematic collection of detailed
quantitative and categorical phenotypic data on each proband allows us to quatitate the
similarity within comparable de novo genotypes (e.g. ARID1B heterozygous loss of function)
compared to radomly chosen groups of probands. Such analyses will allow us to develop
rational approaches to naming entities and grouping genotypes with common phentoypic
effects. In this talk I will present some of the initial results using the first 8000 trios analysed
in DDD.
References:
1: Deciphering Developmental Disorders Study.. Prevalence and architecture of de
novo mutations in developmental disorders. Nature. 2017 Feb 23;542(7642):433-438.
doi: 10.1038/nature21062. Epub 2017 Jan 25. PubMed PMID: 28135719.
2: Deciphering Developmental Disorders Study.. Large-scale discovery of novel
genetic causes of developmental disorders. Nature. 2015 Mar 12;519(7542):223-8.
doi: 10.1038/nature14135. Epub 2014 Dec 24. PubMed PMID: 25533962.
3.Wright et al. Genetic diagnosis of developmental disorders in the DDD study: a scalable analysis of genome-
wide research data. Lancet. 2015 Apr 4;385(9975):1305-14 PMID: 25529582
S34
Notes
S35
Electronic health record phenotyping: An emerging science
Peggy Peissig
Marshfield Clinic Research Institute, USA
Accurately characterizing patients into categories representing disease, exposures, or other
medical conditions is critical when conducting patient-related genomic research. Without
rigorous characterization of patients, also referred to as phenotyping, relationships between
exposures and outcomes cannot be assessed, thus leading to non-reproducible study
results or associations. The electronic health record (EHR) contains highly relational and
inter-dependent biological, anatomical, physiological and behavioral observations and facts
that represent a patient’s phenotype. Developing computerized phenotyping algorithms that
use the EHR data is time consuming and requires medical insight, which is based on the
perceptions and past experiences of clinical experts involved in the phenotyping effort. The
result is a serious temporal and informative bottleneck when constructing high quality
generalizable phenotypes for research.
This presentation will highlight research conducted by the Electronic MEdical Record and
GEnomics network (eMERGE), describing the challenges and complexity of the phenotyping
process and lessons learned. In addition, state-of-the-art phenotyping approaches, tools,
algorithm generalizability and data harmonization techniques will be examined. Using this
information, can we reposition these phenotypes to inform the return of genomic results?
S36
Notes
S37
Defining and refining disease nomenclature based on gene-focused curations in the age of genomic medicine.
Courtney Thaxton1, Jennifer Goldstein1, Kathleen Wallace1, Marina DiStefano2, Dane Witmer3, Melissa Haendel4, Ada Hamosh5, Heidi Rehm2,6, Jonathan Berg1
1 Department of Genetics, The University of North Carolina, Chapel Hill, NC, USA; 2 Laboratory for Molecular Medicine, Partners Healthcare Personalized Medicine, Cambridge, MA, USA; 3 Center for Inherited Disease Research, McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University, Baltimore, MD, USA;4 Monarch Initiative, OHSU, Portland, Oregon, USA; 5 McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University, Baltimore, MD, USA. 6 Center for Genomic Medicine, Massachusetts General Hospital, Boston, MA USA.
A name, whether it be John, Earth, or Italian is a term used for identification. Similarly,
diseases, disorders and syndromes are given a name, such as diabetes or Noonan
syndrome, to identify phenotypic features and to distinguish one diagnosis from another.
Historically, nosology has approached defining disease entities based on the presentation of
phenotypic features, and often the resulting disease nomenclature has been either
eponymously based, such as Marfan syndrome and Pompe disease, or more recently
acronym based for the presenting phenotypic features, such as MELAS or CHARGE.
Presently, due to rapid advancements in sequencing technologies and genomic medicine,
the underlying genetic etiology of many disorders have been identified. Thus, it has
materialized that several disease entities may need systematic reclassification and re-
categorization based on lumping and splitting guidance that accounts for the genetic basis,
as well as refinement and/or defining of the disease nomenclature. This circumstance poses
several questions, including: (1) how should one assign disease nomenclature for any
disease entity; (2) how should one refine the disease nomenclature for a lumped disease
entity that includes an eponymous name; (3) what is the potential impact of changing
disease nomenclature to physicians and patients? As a cooperative effort, the ClinGen
Lumping and Splitting Working Group has engaged OMIM and Monarch Initiative,
nosological and ontological authorities respectively, in assembling criteria not only to
determine when to lump or split for gene-based curations, but also in providing guidance for
naming lumped disease entities. As part of the effort, we will survey clinicians, researchers,
diagnosticians, and patients to determine naming schema that may be readily accepted and
adopted by the greater community. Here, we will present some of the questions, guidance
and examples, as well as engage the audience in a live survey and demonstrate the
responses.
S38
Notes
S39
Exome sequencing of 506 parental/fetal trios with structural abnormalities revealed by ultrasound in the UK Prenatal Assessment of Genomes and Exomes (PAGE) project
Dominic J McMullan [1], Jenny Lord [2], Ruth Eberhardt [2], Gabriele Rinck [2], Sue Hamilton [1], Liz Quinlan-Jones [3], Lucy Jenkins [4], Richard Scott [4], Denise Williams [1], Mark Kilby [3], Eamonn Maher [5], Lyn Chitty [6], Matthew Hurles [2]
[1] West Midlands Regional Genetics Service, Birmingham Women’s and Children’s NHS Foundation Trust, UK, [2] Wellcome Trust Sanger Institute, Hinxton, Cambridge, UK, [3] Department of Fetal Medicine, Birmingham Women's and Children’s Hospital UK, [4] NE Thames Regional Genetics Service, Great Ormond Street Hospital, London, UK, [5] Department of Medical Genetics, University of Cambridge and Cambridge NIHR Biomedical Research Centre UK, [6] Genetics and Genomic Medicine, UCL Institute of Child Health and Great Ormond Street Hospital NHS Foundation Trust, London, UK.
PAGE aims to apply whole exome sequencing (WES) to 1000 trios recruited in the UK-NHS
to identify pathogenic variation underlying heterogeneous fetal structural abnormalities
detected by ultrasound scan (USS). Whole genome sequencing (WGS) is being carried out
on a proportion of cases with complex phenotypes which are negative by WES. Trio WES is
conducted after resolution of pregnancy if conventional testing (QF-PCR, chromosomal
microarray and/or targeted single/panel gene testing) fails to establish a definitive diagnosis.
Genetic variants are triaged via a stringent clinical filtering pipeline established for the UK
Deciphering Developmental Disorders (DDD) project using a gene panel adapted iteratively
from the DDG2P gene panel throughout the course of the project. Potentially pathogenic
variants are assessed and classified by a UK-wide multidisciplinary clinical review panel
(CRP), technically validated in NHS accredited labs and reported back to Clinical Genetics
units and families where appropriate.
From 506 trios so far reviewed, diagnostic variants were identified in 41 cases (8∙7%).
Diagnostic yield varies considerably by phenotypic class, with multisystem phenotypes
showing the highest yield (16%). The majority of variants are SNVs/indels which would
escape targeted detection by conventional testing. When compared to a null model based on
triplet mutation rate, an excess of de novo mutation is observed, more pronounced in known
dominant genes (such as KMT2D). Further analysis is predicted to identify new gene and
mechanistic associations underlying observed phenotypes as more samples are processed.
PAGE aims to catalyse responsible adoption of WES and potentially WGS into routine
prenatal clinical diagnostics and lessons learned will be assimilated into the planned
introduction of WES for such referrals in the NHS in England in 2018/2019.
S40
Notes
S41
Reappraisal of reported genes for sudden arrhythmic death: An evidence-based
evaluation of gene validity for Brugada syndrome
Michael Gollob
University of Toronto, Canada
Clinical validity of gene-disease associations is critical for the accurate application of genetic
testing in patient care. However, evidence-based assessment of clinical validity of gene-
disease associations is not always considered prior to inclusion on genetic testing panels.
Brugada syndrome (BrS) is an arrhythmia syndrome with a risk of sudden death. More than
20 genes have been reported to cause BrS and are routinely assessed on genetic testing
panels.
This presentation will review the design and results of a comprehensive gene curation
evaluation of reported gene-disease associations in Brugada Syndrome on behalf of the
ClinGen Channelopathy Working Group.
S42
Notes
S43
Curating clinically relevant transcripts for the interpretation of sequence variants
Marina T. DiStefano1, Sarah E. Hemphill1, Brandon J. Cushman1, Mark J. Bowser1, Elizabeth Hynes1, Andrew R. Grant1, Rebecca K. Siegert1, Andrea M. Oza1, Michael A. Gonzalez2, Sami S. Amr1,3, Heidi L. Rehm1,3,4,5, and Ahmad N. Abou Tayoun2
1Laboratory for Molecular Medicine, Partners Healthcare Personalized Medicine, Cambridge, MA, USA. 2Division of Genomic Diagnostics, The Children's Hospital of Philadelphia, The University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA. 3Department of Pathology, Brigham & Women’s Hospital, Harvard Medical School, Boston, MA USA 4Center for Genomic Medicine, Massachusetts General Hospital, Boston MA USA 5The Broad Institute of MIT and Harvard, Cambridge, MA, USA
Variant interpretation depends on defining biologically relevant transcripts to accurately
annotate the impact of variation on gene function. We have developed a systematic strategy
for designating primary transcripts for variant interpretation and applied it to 109 hearing
loss-associated genes. Genes were divided into 3 categories. Category 1 (C1) genes (N=38)
had a single transcript, Category 2 (C2) (N=33) had multiple transcripts, but a single
transcript sufficiently represented all exons, and Category 3 (C3) genes (N=38) had multiple
transcripts with unique exons. Transcripts were curated with respect to gene expression
reported in the literature and the Genotype-Tissue Expression (GTEx) Project. In addition,
exonic loss-of-function (LoF) variants with a frequency over 0.3% were queried from the
Genome Aggregation Database (gnomAD). All variants classified as pathogenic or likely
pathogenic in ClinVar or as DM in the Human Gene Mutation Database were pulled and
evaluated for each exon. These data were used to classify exons. "Clinically significant"
exons lacked high frequency LoF variants or were supported by literature, "Uncertain
significance" exons were spliced of out major transcripts, had no data in the literature, or,
contained one high frequency LoF variant, and "Clinically insignificant" exons had non-
supporting expression data or had multiple high frequency LoF variants. Interestingly, 7% of
all exons were of "uncertain significance", yet contained >124 variants reported as clinically
significant, questioning their accurate interpretation. Finally, we used exon-level next
generation sequencing quality metrics generated across exome samples analyzed at two
clinical labs to identify a total of 43 exons in 20 different genes that had inadequate coverage
and/or homology issues which may lead to missed or false variant calls. We have
demonstrated that transcript analysis plays a critical role in accurate variant interpretation.
S44
Notes
S45
Implementation of gene curation in a clinical laboratory setting
Dr Alison Coffey, Krista Bluske, Anjana Chandrasekhar, Julie P. Taylor, Revathi Rajkumar, Ursula Webber, David R. Bentley, Ryan J. Taft, Denise L. Perry
Illumina Clinical Services Laboratory, Illumina Inc., San Diego CA, USA
We have recently launched the Illumina Clinical Services Laboratory (ICSL) Gene Curation Programme (GCP) based on the ClinGen Framework to support our clinical whole genome sequencing screen (cWGS) for suspected rare and undiagnosed genetic diseases and build a resource of curated gene-disease associations (GDAs) with an initial focus on genes associated with childhood onset disorders. The GCP aims to determine if GDAs meet criteria for clinical reporting or are candidates for research and submission to GeneMatcher. Additionally, the process includes investigation of disease mechanism, and production of both a gene and disease description, enhancing our knowledge base and assisting in variant curation. The workflow consists of an initial clinical review of potential GDAs which are passed to curators, who are experts in the framework and curation process. Completed curations are reviewed first by the scientific team and subsequently by the clinical team. The clinical review is an important part of the process for multiple GDAs associated with a single gene. By defining phenotypic overlap between related disorders and giving insight into disease management, the clinical expertise enables us to determine if multiple associations should be described individually on a clinical report, or together as a gene-related spectrum disorder. At present we have 189 genes with 254 GDAs in our pipeline. To evaluate concordance, sixteen GDAs curated by ClinGen were completed by ICSL curators who were blinded to the reported classification. All curations were concordant with the clinical validity awarded. Forty-one GDAs curated by the BabySeq Project were also compared. Twenty-eight (68%), classified as "Definitive", were fully concordant. Of the 13 discordant GDAs, ten classifications were upgraded, with additional evidence. Three GDAs were downgraded following identification of new evidence suggesting a classification of "Conflicting". As part of the expansion of our secondary findings analysis, we are curating all GDAs associated with the ACMG59 genes and other clinically actionable genes. We have recently extended our cWGS screen to include interpretation of SNVs in the mitochondrial genome and are developing a framework for the curation of mitochondrial GDAs to aid in the assessment of variant pathogenicity. All of our gene curation data will be shared with the ClinGen Knowledge Base.
S46
Notes
S47
Assessing the strength of evidence for genes implicated in fatty acid oxidation disorders using the ClinGen Clinical Validity Framework
Jennifer McGlaughon1, Heather Baudet1, Stephanie Crowley1, Gregory Enns2, Annette Feigenbaum3, C. Lisa Kurtz1, Elaine Lyon4, Marzia Pasquali4, Justyne Ross1, Ozlem Senol-Cosar5, Wei Shen4, Kathleen Wallace1, Meredith Weaver6, Rong Mao4
1University of North Carolina at Chapel Hill, NC; 2Department of Pediatrics and Pathology, Stanford University, Stanford, CA, USA; 3Department of Pediatrics, University of California San Diego; Rady Children’s Hospital, San Diego, CA, USA; 4ARUP Institute for Clinical and Experimental Pathology; Department of Pathology, University of Utah, Salt Lake City, UT, USA; 5Laboratory for Molecular Medicine, Partners Healthcare Personalized Medicine, Cambridge, MA, USA; 6American College of Medical Genetics and Genomics, Bethesda, MD, USA
The mitochondrial fatty acid β-oxidation pathway plays an important role in energy
production during times of catabolic stress. Fatty acid oxidation disorders (FAODs) are
autosomal recessive conditions caused by defects in the pathway, which comprises at least
20 individual steps. Newborn screening (NBS) based on acylcarnitine profiling of blood spots
is used to identify FAODs in newborns, followed by biochemical and DNA testing to confirm.
The ClinGen Inborn Errors of Metabolism Working Group was established to create a
knowledge base of genes and variants relevant for metabolic diseases and genomic
medicine. As part of this effort, the FAO Subgroup was created to examine the strength of
association between FAODs and genes implicated in these disorders using the ClinGen
Clinical Validity Framework (Strande & Riggs et al., 2017). This process involves the
curation and evaluation of publicly available evidence to assign the strength of a gene-
disease association into one of the following classifications: Definitive, Strong, Moderate,
Limited, No Reported Evidence, or Conflicting Evidence Reported. The FAO Subgroup
developed a comprehensive list of 23 gene-disease associations for evaluation using the
framework. The process for assigning a clinical validity classification is as follows: (1)
Curations are performed by biocurators and a provisional classification is assigned. (2) Two
disease experts review the curation and indicate whether they agree with the provisional
classification. (3) If the experts agree, the classification is finalized on a conference call with
the entire group. (4) If the experts disagree, the curation is discussed and a final
classification is reached. Thus far, we have completed the curation of more than 8 gene-
disease associations. We will present the results of our curation effort to date.
S48
Notes
S49
Balancing the sensitivity and specificity of variant classification for healthy
populations
Peter Kang1, Samuel Cox1, Krista Moyer1, Rebecca Mar-Heyming1
1. Counsyl, 180 Kimball Way, South San Francisco, CA 94080
Historically, diagnostic genetic testing has been inherently skewed towards minimizing false
negatives, with a focus on resolving the pathology of an appreciable patient phenotype. In
contrast, when screening healthy individuals, a bias towards minimizing false positives is
preferable in order to avoid suggesting unnecessary clinical management in the absence of
sufficient evidence for disease. Such opposing prioritization of either sensitivity or specificity
has inevitably contributed to the incidence of conflicting gene variant classifications, despite
ACMG guidelines suggesting a 90% certainty threshold for assigning pathogenic
designations.
In genetic carrier screening of healthy individuals, variant population frequencies are an
important tool for the re-evaluation of variants deemed disease-causing from diagnostic
testing. Researchers have suggested that variants whose frequency exceeds overall disease
incidence, or the frequency of the most common known pathogenic variant for that disease,
can be appropriately classified as benign. As more patients with genetic disease are studied,
affected patients who carry a certain rare benign variant are more likely to be identified.
However, some publicly available pathogenic classifications appear to disproportionately rely
on reports of affected patients in the literature with the relevant allele. At Counsyl we have
determined a number of such variants to be insufficiently enriched in cases compared with
the general population to warrant a deleterious classification and have subsequently
classified them as benign or of uncertain significance.
For example, NM_000053.3(ATP7B):c.122A>G(N41S) has been reported in a total of eight
published cases affected by Wilson disease which seems to support a pathogenic
classification. Indeed, ClinVar submissions from five separate laboratories classify N41S as
likely pathogenic. However, we find the frequency in cases to be insufficiently different from
a relatively high frequency for this variant in population databases (54/126572 in
Europeans). After the assessment of all available evidence, we have classified N41S as a
variant of uncertain significance (VUS). Other similar examples will be discussed.
To further inform classifications, particularly where evidence places variants on the cusp of
pathogenic versus uncertain designations, we have also developed simulations that
incorporate curated estimates of total reported case counts for individual diseases. These
developments, together with data obtained through alternative methods, will additionally be
presented.
S50
Notes
S51
Genetic cascade screening for Familial Hypercholesterolemia: a national
cardiovascular disease prevention programme
Joep Defesche
Department of Clinical Genetics, Genome Diagnostics, Academic Medical Centre at the
University of Amsterdam, Amsterdam, The Netherlands.
Familial Hypercholesterolemia (FH) in its heterozygous form, with a prevalence of 1 in 250
persons in most European countries, is the most frequent genetic metabolic disorder. If left
untreated, FH poses a very high risk for premature cardiovascular disease and death. Highly
effective treatment for FH resulting in complete normalization of cholesterol levels and
cardiovascular risk, is available.
In The Netherlands a genetic diagnostic service is available for FH and other genetic
dyslipidaemias, covering the whole country. This service comprises the establishment of a
genetic diagnosis in patients with a clinical suspicion of FH and the subsequent genetic
cascade screening of family members of a molecularly characterised patient with FH.
The genetic cascade screening programme was initiated by our institute in 1994 and is still
ongoing. From the year 2000 until 2014, the programme was run under auspices and
finance of the Ministry of Public Health.
On a yearly basis, DNA analysis is performed on about 2000 clinical index cases and
approximately 4500 family members of index cases are actively contacted, visited at home
or at their place of work for blood sampling and expansion of the family’s pedigree. Blood
samples are analysed for the family-specific mutation. This results in the identification of
about 1800 new cases of FH per year.
Up to December 2016, a total 63.384 persons participated in the programme resulting in the
identification of 29.463 persons with genetically confirmed FH.
During the course of the genetic cascade screening programme for FH a substantial amount
of studies have been undertaken and published, addressing virtually all issues associated
with such a large scale population screening: efficiency of the programme, costs-
Next generation sequencing has dramatically changed the pace of gene discovery in rare
disorders, with hundreds of molecular basis identified each year through a hypothesis free
approach. This powerful tool allows the reanalysis of data in the light of new publications.
This study presents the experiment of a French regional center performing solo clinical
whole-exome sequencing for rare disorders with congenital anomalies and/or intellectual
disability. Raw data of the non-positive results were reanalyzed every year on an updated
bioinformatic pipeline and a double clinico-biological interpretation was realized. The
diagnostic yield of the first analysis of the 416 patients’ data was 25%. Prospective
reanalysis allowed the resolution of 46 additional cases, raising the yield to 36%. 27 cases
were resolved through a strict diagnostic approach based on the ACMG guidelines; 19
through a translational research based on international data sharing and reverse
phenotyping. The mean number of etiological tests prior to WES significantly decreased,
highlighting the economic interest of this strategy. Although the reanalysis do not lead to
technical overcost, it is time-consuming and appears difficult to systematize in a regional
center. This work underscores the considerable interest of periodically reanalysis of WES
data and of a translational integrated organization from diagnosis to research.
S56
Notes
S57
Implementation of a whitelisting approach to make additional diagnoses of single-gene developmental disorders in whole exome trios
Panayiotis Constantinou(1,2), Caroline Wright(1,3), David FitzPatrick(4), Helen V Firth(1,2), Matthew E Hurles(1), on behalf of the Deciphering Developmental Disorders Study
(1)Wellcome Sanger Institute, Hinxton, UK; (2)East Anglia Regional Genetics Service, Addenbrooke's Hospital, Cambridge, UK; (3)University of Exeter Medical School, Institute of Biomedical and Clinical Science, Exeter, UK; (4)MRC Human Genetics Unit, Edinburgh, UK
The UK-wide Deciphering Developmental Disorders (DDD) study has utilised trio-whole
exome sequencing (WES) to search for the causes of severe developmental disorders in
over 13,000 children. Extensive phenotypic information and WES data has been collated for
these children and both their parents, where available. Diagnostic yields of up to 40% have
been achieved with iterative reanalysis of the first 1,000 patient-parent trios. The
bioinformatics pipeline for analysing potentially clinically relevant variants involves
automated variant annotation, filtering and prioritisation using a curated list of genes causing
developmental disorders (DDG2P) with defined allelic requirements and mutational
consequences.
A potential additional source of clinically relevant variants which has yet to be incorporated
into the pipeline is that of aggregated, publicly available reports linking genetic variation and
phenotypes, backed up by supporting evidence. The largest archive of such reports is the
ClinVar database, hosted by the National Center for Biotechnology Information, NIH, USA.
At the time of writing, ClinVar has over such 600,000 records of around 380,000 unique
variants.
We generated a "whitelist" of known pathogenic variants in ClinVar affecting DDG2P genes
and used this to filter against the sequencing output of over 8,000 patient-parent trios in the
DDD study. We will present a series of additional potential diagnoses arising from this
approach as well as highlighting issues such as possible incomplete penetrance and
questionable pathogenicity of variants in publicly available archives such as ClinVar.
S58
Notes
S59
Scaling the resolution of sequence variant interpretation discrepancies in ClinVar
Steven M. Harrison1, Jill S. Dolinsky2, and Heidi L. Rehm1,3 and ClinGen Sequence Variant Inter-Laboratory Discrepancy Resolution Working Group
1Partners HealthCare Laboratory for Molecular Medicine, Cambridge, MA, USA; 2Ambry Genetics, Aliso Viejo, CA, USA; 3The Broad Institute of MIT and Harvard, Cambridge, MA, USA
Sharing data in ClinVar provides open access to variant classifications from many clinical
laboratories. While the majority of classifications agree, ClinVar has shed light on the
important issue of interpretation differences between laboratories, providing an opportunity
to resolve differences and positively impact patient care. A recent ClinVar study found that
81% of variants had concordant interpretations while 89% reached a majority consensus
(agreement in classification of ≥ ⅔ of submitters), suggesting that for a subset of
discrepancies, the majority of submitters agree with an outlier interpretation(s) accounting for
the discrepancy (PMID: 28569743). Additionally, a pilot project from ClinGen's Sequence
Variant Inter-Laboratory Discrepancy Resolution team focusing on four clinical laboratories
found that 53% of interpretation differences were resolved by either updating ClinVar with
current internal classifications or reassessment of an older interpretation with current
classification criteria (PMID: 28301460). With these findings in mind, our working group
expanded to include 41 clinical laboratories and prioritized variants with outlier
interpretations. Comparison of interpretations from 41 clinical laboratories identified 24,445
variants interpreted by ≥2 clinical laboratories (April 2017). The majority of classifications
were concordant 84.6% (20,677 variants). Only 2.7% (650 variants) of variants were
medically significant differences (MSDs) with potential to impact medical management
[pathogenic (P/LP) versus other (VUS/LB/B]. Of the MSDs with ≥3 interpretations (244
variants), 87.6% (213 variants) reached a majority consensus, thus allowing for identification
of outlier submissions most in need of reassessment. Laboratories with outlier interpretations
were sent a custom report and encouraged to update ClinVar with current classifications and
reassess remaining conflicts. Laboratories have returned results for 204 variants, of which
62.3% (127 variants) were resolved by this process. The project has now expanded to all
clinical laboratories submitting to ClinVar (108 submitters) and outlier reports have been sent
to 22 of the 32 laboratories with ≥2 outlier interpretations (March 2018). This process adds to
the value of ClinVar and will help the community move toward more consistent variant
interpretations which will improve the care of patients with, or at risk for, genetic disorders.
S60
Notes
S61
GenomeConnect: Sharing individual level data through patient registries
Juliann M. Savatt1, Danielle R. Azzariti2, W. Andrew Faucett1, Steven M. Harrison2,
Jennifer Hart3, Melissa J. Landrum3, David H. Ledbetter1, Vanessa Rangel Miller4, Emily
Palen1, Heidi L. Rehm2,5,6,7, Jud Rhode4, Erin Rooney Riggs1, Jo Anne Vidal4, Christa
Lese Martin1 on behalf of the Clinical Genome (ClinGen) Resource
1Geisinger, Danville, Pennsylvania, USA; 2Laboratory for Molecular Medicine, Partners
Personalized Medicine, Boston, Massachusetts, USA; 3 National Center for Biotechnology
Information, Bethesda, Maryland, USA, 4Invitae, San Francisco, California, USA; 5The
Broad Institute of Harvard and MIT, Cambridge, Massachusetts, USA; 6Harvard Medical
School, Boston, Massachusetts, USA; 7Department of Pathology, Brigham & Women’s
Hospital, Boston, Massachusetts, USA
Participants in GenomeConnect (GC), the Clinical Genome Resource (ClinGen) patient
registry, consent to have their genetic and health information de-identified and shared with
approved databases. The goal of this data sharing is to increase availability of genotypic and
phenotypic information to aid in variant interpretation and improve patient care. Health
information is collected via participant completed surveys and genomic data is derived from
participants’ genetic test reports. De-identified data is shared with NCBI’s ClinVar; variants in
genes of uncertain significance are submitted to GeneMatcher. Participants can be re-
contacted to request additional information and provide the option of connecting with
clinicians or researchers. To assess the impact of patient data sharing, we reviewed GC
ClinVar submissions. Of 732 sequence variants, 47.9% (n=351) were not submitted
previously, demonstrating the importance of patients as a genomic data source. Of the
previously reported variants, only 60.9% (n=232/381) had been previously submitted by the
participant’s reporting laboratory. For 13 variants, the participant’s report was outdated
compared to the laboratory’s current ClinVar entry. GC provides participants the option to
receive such classification updates. To date, 96.6% (n=704/729) of participants that have
updated their preferences opted to receive updates. Of variants previously submitted by a
laboratory other than the reporting institution, 41.3% (n=107/259) had a difference in major
category classification between the participant report and another submission. Although this
information will not be relayed to participants at this time, GC is working with the ClinGen
Sequence Variant Discrepancy Resolution group to encourage laboratories to address these
discrepancies. Moving forward, ClinGen plans to increase patient data sharing by also
partnering with external registries, and advocacy groups. By engaging patients in data
sharing, ClinGen and GC contribute information to the public knowledge base, benefiting
both patients and the genomics community.
S62
Notes
P1
Poster Presentations
Trusted Variant eXchange: a database for secure sharing of variant classifications between trusted partners
Sharmini Alagaratnam, Tony Håndstand, Valtteri Wirta
(SA) Life Sciences Programme, Group Technology and Research, DNG VL, Høvik, Norway; BigMed (TH) Department of Medical Genetics, Oslo University Hospital, Norway; BigMed (VW) Clinical Genomics facility, Science for Life Laboratory, Karolinska Institutet, Stockholm; BigMed
Guidelines aim to standardize the classification of genetic variants for rare diseases and
cancer. However, adherence to guidelines can vary greatly within a single healthcare system
or country, leading to discordance in variant classification. International classification
databases such as ClinVar can support the variant interpretation process, but issues of data
incompleteness and inconsistency remain. To address this, we have developed the Trusted
Variant eXchange, or TVX.
Requirements for TVX were collected and collated through a series of workshops and
feedback cycles with our clinical partners through BigMed, a research project funded by the
Norwegian Research Council. The database was designed with the following in mind:
scalability, for varying data volumes; and adaptability, for evolving technological and
regulatory needs. The database was built using components and microservices, based
around blob and table storage, queues and functions.
TVX database facilitates sharing of evidence-based classification of interpreted variants
between trusted clinical diagnostic partners, focusing on data quality and conflict reporting.
After authentication, partners can submit and search for variants with their associated
classifications through a secure API. Entering all new variant classifications builds up a
collective, high-quality knowledge base with high transparency and traceability, with the
ability to share further to international databases. Discordances in classification are flagged
and resolution facilitated through communication with the relevant partners.
Harmonization of variant classification is a priority for multi-site healthcare systems where
equal access to quality healthcare is a goal. TVX enables such harmonization while
continuously curating and accumulating expertise.
P2
Blockchain-based Framework to Support Data Sharing in Clinical Informatics
Faisal Albalwy, Prof Andrew Brass, Dr Angela Davies
School of Computer Science, Kilburn Building, University of Manchester, Oxford Road, Manchester M13 9PL, UK Division of Informatics, Imaging and Data Sciences, Stopford Building, University of Manchester, Oxford Road, Manchester M13 9PL, UK; Division of Evolution & Genomic Sciences, Stopford Building, University of Manchester, Oxford Road, Manchester M13 9PL, UK
The advent of fast and effective genome sequencing technologies has led to a step change
in the diagnosis of rare genetic conditions. Because these conditions are rare, it is very
important that genome centres can share data to make the best use of this information in
order to improve diagnosis and treatment. However, genomic data sharing has proven to be
very difficult to achieve in practice. Genetic data is highly sensitive personal information.
Some genetic variants are rare enough that revealing them could be considered identifying
information; therefore, these data must be subject to strict data privacy rules. In addition,
concerns regarding data governance rules, security and a lack of established standards
have become obstacles to genomic data sharing. One possible option to support the sharing
of sensitive genomic data is to store it in a cloud storage environment and provide secure
access to interested parties. However, giving sensitive data to third parties raises security
and privacy concerns. Another option is to use Blockchain technology. Given its security and
privacy advantages, and its principle feature of decentralising data management, Blockchain
technology has the potential to revolutionise the sharing of, and access to, sensitive genomic
data. This project will investigate the use of Blockchain technology to see whether it can be
effective in this area. It will also use data sharing in clinical informatics as a use case to
characterise larger challenges surrounding data sharing in the context of complex
governance infrastructures such as healthcare.
P3
The Recreational Genome: Genetic Counseling Following Direct to Consumer DNA Testing
Direct to consumer (DTC) genetic testing is controversial but growing. The market has tripled
since 2012 and is expected to be US$340M by 2022. While predominantly for ancestry,
many companies provide customers access to the raw data and the option to evaluate it for
health risks using third party software. In addition, several companies now offer whole
exome sequencing (WES) directly to consumers, sometimes requiring the approval of a
physician, who is often unprepared to interpret the results. There is a great need to
contextualize and interpret information for users and many companies offer customers the
option to seek a consultation with a genetic counselor. While genomic test interpretation and
counseling is within the scope of practice for genetic counselors; few do it and there are no
established protocols or practice guidelines.
An independent genetic counselor has provided raw data interpretation for 22 cases. 5
cases for SNP-based genotyping by 23 & Me followed by Promethease or other third party
software; 17 cases for WES through Genos. All clients sought consultations post-test and
were counseled over the phone or using a web-based screen sharing platform. Clients were
from the US (19), England (1), Hungary (1) and Australia (1). There were 8 (36%) males and
14 (64%) females. Ancestry was European (20), Indian (1), and unknown/adopted (1).
Clients' ages ranged from 23 to 74y (median age 52y). All of the clients had university
degrees: 11 were in a scientific or medical field, 7 of those had doctoral degrees. Reasons
for seeking DTC testing included: curiosity and the desire to know at risk conditions (n=10,
45%); to determine a genetic cause for personal medical issues (n=6, 27%); concern about
a family history of cancer (n=3,14%) or dementia (n=3,14%). Consultations included: review
of benefits and limitations of test; limitations of interpretation and the potential for change
over time; basic genetics (inheritance and GWAS); absolute vs relative risk; how to navigate
a report; how to evaluate variants; discussion of most relevant variants. None had a
pathogenic variant in the ACMG 59. Two (9%) clients requested a second session to further
discuss results. Recommendations for confirmatory clinical testing were made in four cases
(18%). Follow-up genetic cancer risk assessment, based on family history, was
recommended and scheduled in one case (5%).
P4
Variant pathogenicity evaluation through the community-driven Inherited Neuropathy Variant Browser
Dana M. Bis (1), Cima Saghira (1), David Stanek (2), Alleene Strickland (1), David N. Herrmann (3), Mary M. Reilly (4), Steven S. Scherer (5), Michael E. Shy (6), Inherited Neuropathy Consortium, Stephan Züchner (1)
(1) Department of Human Genetics and Hussman Institute of Human Genomics, University of Miami, Miami, USA (2) DNA Laboratory, Department of Paediatric Neurology, 2nd Faculty of Medicine, Charles University in Prague and University Hospital Motol, Prague, Czech Republic (3) Department of Neurology, University of Rochester, Rochester, New York, USA (4) MRC Centre for Neuromuscular Diseases, UCL Institute of Neurology, Queen Square, London, UK (5) Department of Neurology, University of Pennsylvania, Philadelphia, Pennsylvania, USA (6) Department of Neurology, University of Iowa, Iowa City, Iowa, USA.
Charcot-Marie-Tooth disease (CMT) is an umbrella term for inherited neuropathies affecting
an estimated 1 in 2500 people. Over 120 CMT and related genes have been identified and
clinical gene panels often contain more than 100 genes. Such a large genomic space will
invariantly yield variants of uncertain clinical significance (VUS) in nearly any person tested.
This rise in number of VUS creates major challenges for genetic counseling. Additionally,
fewer individual variants in known genes are being published as the academic merit is
decreasing, and most testing now happens in clinical laboratories, which typically do not
correlate their variants with clinical phenotypes. For CMT, we aim to encourage and facilitate
the global capture of variant data to gain a large collection of alleles in CMT genes, ideally in
conjunction with phenotypic information. The Inherited Neuropathy Variant Browser provides
user-friendly open access to currently reported variation in CMT genes. Geneticists,
physicians, and genetic counselors can enter variants detected by clinical tests or in
research studies in addition to genetic variation gathered from published literature, which are
then submitted to ClinVar bi-annually. Active participation of the broader CMT community will
provide an advance over existing resources for interpretation of CMT genetic variation.
P5
Challenges of curating and classifying variants detected in healthy populations, including in the ACMG 59
Nicole J. Burns, Alison J. Coffey, Amanda R. Clause, David R. Bentley, Ryan J. Taft, Denise L. Perry
Illumina Clinical Services Laboratory, Illumina, Inc., San Diego, CA, USA
From 2012 to 2018, the Illumina Clinical Services Laboratory offered its physician-ordered
clinical whole genome sequencing (cWGS) predisposition screen for presumably healthy
adults. This screen detected single nucleotide variants and small insertions and deletions in
exons and within 15 bp of splice site boundaries in approximately 1,700 genes associated
with Mendelian disorders, including the 59 genes defined as actionable by the American
College of Medical Genetics and Genomics (ACMG).
More than 150,000 unique variants were identified among approximately 2,000 screened
individuals. This considerable number of variants presented challenges for providing timely
and accurate clinical classifications. We developed an efficient variant curation workflow that
utilized an autocategorization algorithm to calculate a score that allowed automatic
classification of variants too common to contribute to disease. Variants not autocategorized
were subject to manual curation. In addition to variant and disease considerations, we
employed internally developed criteria to ensure consistent classification of clinically
significant variants. We utilized the classification categories suggested by ACMG plus
another category, variant of unknown significance-suspicious, which is used for variants with
insufficient data to warrant a likely pathogenic classification in the context of a reportedly
healthy individual but with some evidence suggesting a contribution to disease.
Approximately 80% of screened individuals were carriers for at least one autosomal
recessive disorder. Clinically significant variants in genes associated with autosomal
dominant disorders were identified in roughly 20% of individuals. While this test focused on
only a portion of characterized genes, most individuals screened received findings of
medical significance, demonstrating the value of genetic screening in healthy populations.
Around three percent of individuals carried clinically significant variants in one of the ACMG
59. Strict application of ACMG guidelines is difficult when screening reportedly healthy
individuals, as more evidence is needed to reach clinical significance in the absence of a
disease phenotype. We are using the knowledge gained from curating thousands of variants
identified in healthy individuals to improve the interpretation of variants identified in the
context of secondary findings analysis included as part of our rare and undiagnosed genetic
disease cWGS test. Further discussion is necessary to ensure that clinical laboratories are
aligned in the application of classification guidelines.
P6
Curating mitochondrial variants and genes identified through clinical whole genome sequencing
Ms Nicole Burns, Alison Coffey, Krista Bluske, Aditi Chawla, Alka Malhotra, Julie Taylor, David Bentley, Ryan Taft, Denise Perry
Illumina Clinical Services Laboratory, Illumina Inc., San Diego, CA, USA
Mitochondrial variants are routinely assessed by Sanger sequencing, quantitative PCR, and
long-range PCR followed by massively parallel sequencing. These techniques are often
performed in addition to other molecular investigations, adding complexity to test cadence
and clinical care. The Illumina Clinical Services Laboratory offers clinical whole genome
sequencing (cWGS) intended to identify the underlying cause of a genetic condition through
a single test. The current test definition targets single nucleotide variants (SNVs), insertions
up to 31 base pairs, deletions up to 27 base pairs, and copy number variants greater than 10
kilobases in the nuclear genome and was recently extended to include SNVs in the
mitochondrial genome.
We have developed a protocol to curate mitochondrial SNVs. Evidence, including
information on disease associations, presence in control populations, and in silico
predictions, is gathered from publicly available resources, such as MITOMAP, MSeqDR,
MitImpact, and published literature. Careful evaluation of functional evidence, for example
transmitochondrial cybrid studies or evidence of cosegregation of the variant in a disease
tissue, is also required to determine variant pathogenicity. Variants are classified as
pathogenic, variant of unknown significance, or likely benign, based on the criteria described
in Wang et al. 2012 (PMID: 22402757). Using our protocol, we have curated multiple
mitochondrial variants identified through our cWGS workflow, including m.3243A>G, a
known pathogenic variant in MT-TL1, a gene associated with a disease with strong
phenotypic overlap with the proband's reported clinical presentation, which was detected in a
19 year-old female showing symptoms of mitochondrial leukoencephalopathy, seizures and
choreoathetosis.
In addition to mitochondrial variants, as part of our Gene Curation Programme, we are
developing a protocol to curate mitochondrial gene-disease associations. Our preliminary
investigations suggest that the current ClinGen framework is unsuitable for the curation of
mitochondrial genes because mitochondrial variants show a non-Mendelian, maternal
pattern of inheritance complicated by factors such as heteroplasmy and tissue-specific
threshold effects. Experimental evidence from model systems is rare, and gene-level
experimental data are limited. Therefore, the classification of many mitochondrial gene-
disease associations will rely on variant-level evidence. Given there are currently no
published standards for mitochondrial variant and gene curation, our protocol will contribute
to the development of guidelines for the curation community.
P7
Strategies for Classifying Diverse Mobile Element Insertions
Raymond C. Chan, Jeroen Van Den Akker, Robert O’Connor, Lawrence Hon, Anjali Zimmer, Alicia Y. Zhou, Jack Ji, Scott Topper
Color Genomics
Mobile elements are pervasive in the human genome, comprising an estimated 45% of the
repetitive regions of the human genome (Burns and Boeke 2012). The Alu, Long
Interspersed Element-1 (LINE1) and the composite SINE-VNTR-AluS (SVA) elements are
estimated to generate one new insertion per 20-200 human births (Burns and Boeke 2012).
Collectively, over 100 disease and cancer cases have been attributed to these three
retrotransposons (Burns and Boeke 2012, Hancks 2016).
Accurate classification of a transposon insertion depends on a detailed characterization and
description of the event. It is generally not sufficient to simply recognize that an insertion
exists; one needs to understand the exact nature of the DNA change in the frame of the
transcript, and the consequence of that change on transcription, translation, and resultant
gene product. It can be challenging to assess these aspects, and common descriptions of
mobile elements routinely neglect to assess or provide this information. Mobile elements are
variable in the following ways: type, sequence, orientation, and extent of inserted element
(complete or truncated). We suggest that this information should routinely be interrogated
and communicated to provide a fundamental rationale for classification.
Since early 2017, we have identified 18 unique retrotransposition events in our NGS-based
hereditary cancer panel, encompassing the three major classes of non-LTR
retrotransposons. Here we present the strategies we use to characterize and classify
insertion events: by leveraging the molecular mechanism for retrotransposition and
understanding how the breakpoints would appear in NGS data, we generate a model for the
transposition event. For example, non-LTR retrotransposons share a common canonical
mechanism for genome integration, involving cleavage and duplication of the integration site
(termed tandem site duplication) and the insertion of the transposon via reverse
transcription. We use these known insertion/cleavage sequence preferences of the
retrotransposon encoded enzyme to define the boundaries of tandem site duplications.
Additionally, we infer the sense/antisense orientation and sequences of the transposon from
signals in the NGS data to generate a model of the insertion that is used to analyze its
potential functional impact. Secondary confirmation of the variant is performed by a variant-
specific PCR approach, and a final assessment is made for classification.
P8
“This is the first I’ve heard about it”: Training the Healthcare Workforce for the
Genomics Era
Olath MYA, Leeding J, Chandratillake GL
Cambridge University Hospitals NHS Foundation Trust on behalf of the East of England
Genomic Medicine Centre, UK
Mainstreaming of genomics will require a genetically literate healthcare workforce. To
assess current awareness and inform education and training strategy, the East of England
Genomic Medicine Centre conducted a Training Needs Assessment survey. Responses
were received from >1000 NHS staff, from 40 hospital Trusts and 25 primary care groups
across the region.
Through quantitative and qualitative analysis, general themes and specific educational
needs emerged. Doctors and healthcare scientists generally received genetics training
during their university education, but few received subsequent professional training. The
majority of nurses & midwives, pharmacists, and allied health professionals reported
receiving no genetics training at any time. Awareness of the high-profile national 100,000
Genomes Project stood at 50%, with the appetite for training in genetics being high.
As part of the national Genomics Education Programme, the survey responses have guided
immediate and longer term awareness and educational activities and strategies in the
region, and nationally, which will be presented.
P9
Variant Interpreter: software for interpretation of clinical cancer genomes
R Keira Cheetham, G Jawahar Swaminathan (on behalf of the BSVI team)
Illumina Cambridge Ltd, Chesterford Research Park, UK
Whole genome sequencing of individual patients is now accessible to the fields of oncology
and rare disease. One challenge brought by this technology is the rapid interpretation of the
millions of variants that can be identified in a single sample. BaseSpace® Variant Interpreter
(BSVI) is cloud-based software which can aid in the interpretation of sequencing data. Data
can be ingested directly from BaseSpace Sequence Hub® or variant call format files (VCFs)
can be uploaded individually or in batches. The integrated Annotation Engine® evaluates the
consequence of all variants (single nucleotide, insertion/ deletions, structural variants and
copy number variants) in a sample and displays them in an interactive grid, allowing
browsing, filtering, review and interpretation of the data. Annotations include COSMIC (the
Catalogue of Somatic Mutations in Cancer), ClinVar, population frequency and DGV
(Database of Genomic Variants), and links out include Ensembl Genome Browser, UCSC
Genome Browser, OMIM (Online Mendelian Inheritance in Man) and IGV (Integrative
Genomics Viewer) for inspection of the sequencing reads. Filters include variant call metrics,
population frequency, consequence, annotations and custom gene or region lists. Variants
are displayed alongside a knowledge base of curated associations (KnowledgeBase), such
as clinical trials and publications. Users can add their own curations to their local
KnowledgeBase, which can be shared between workgroups. The KnowledgeBase reacts to
the tumour type of the sample, highlighting the most relevant associations. Alongside the
interactive variant grid, Variant Interpreter provides both static and dynamic genome-wide
visualisations. These provide a snapshot of the large structural rearrangements and a
means of browsing the data. Mutational signatures and tumour content are both
informatically predicted.
BSVI is designed to fit into existing laboratory workflows. Workflow and case management
tools are built-in. There is an export capability, audit log and bulk actions. BSVI supports
sequencing panels as well as whole genome tumour-normal or tumour-only samples. It can
also be used for germline samples including family-based analysis in rare disease.
BSVI is being developed by Illumina in collaboration with Genomics England and has been
released to the Genomic Medicine Centres (GMCs). BSVI is also available in the cloud via
Amazon Web Services as part of Illumina's BaseSpace® suite of genomic analysis tools at
variantinterpreter.informatics.illumina.com
P10
BaseSpace Informatics Suite - Powering factory scale sequencing via an extensible informatics reference architecture.
Donavan T. Cheng, G Jawahar Swaminathan, R Keira Cheetham
Illumina Inc.
Significant reductions in the cost of genomic analysis have enabled large scale sequencing
initiatives and have spurred efforts to integrate whole genome sequencing (WGS) into the
health care system, in areas such as cancer and genetic disease. Comprehensive analysis
and accurate variant interpretation remain a bottleneck, preventing NGS from being adopted
for routine testing and enabling sample-to-answer paradigms. We present the BaseSpace
Informatics Suite as a scalable and extensible Informatics Reference Architecture, capable
of delivering analysis and interpretation on 1000s of whole human genomes per month. The
Reference Architecture has capabilities for managing samples through the laboratory, for
executing alignment and variant calling workflows using BaseSpace Sequence Hub (BSSH)
and for interpreting variants using BaseSpace Variant Interpreter (BSVI). BSSH uses a
flexible Docker container system to allow users to create and automate pipelines. Open APIs
are available for programmatic access to data by authorized 3rd parties. Utilities like
BaseMount and Command Line Interface (CLI) enable batch processing and automated
workflow kickoff. To facilitate WGS analysis, default workflows are built-in by default,
enabling detection of a wide variety of variants, i.e. SNPs, INDELS, structural variants, copy
number variants, repeat expansions, mitochondrial variants, HLA typing, and regions of
homozygosity. Germline and tumor/normal somatic variant calling are supported, as well as
joint variant calling in small pedigrees. BSVI is a collaborative knowledge sharing tool that
expedites variant annotation, filtering, curation, interpretation and reporting. Key features of
BSVI include aggregate annotations from a broad range of public and private knowledge
bases, interactive visualizations and regulated data sharing across workgroups. Sample
genotypes and phenotypes are held in a big data warehouse, accessible via a high-
performance query engine for returning case/control statistics within seconds. Users can
customize knowledge content and variant annotations via bulk upload. Multiple components
in the reference architecture have been developed in collaboration with Genomics England
as part of the BCIP collaboration. The platform is being released to the Genomic Medicine
Centres (GMCs) to capture feedback and to ensure requirements are met. This informatics
reference architecture is designed for easy replication at other large scale sequencing
initiatives, and will help establish proof points that genome sequencing can be rolled out at
scale for health care systems.
P11
Familial hypercholesterolemia-associated variants submitted to ClinVar - a ClinGen FH effort
Joana Rita Chora1, Michael A. Iacocca2, Marina T. DiStefano3, Alain Carrie4, Tomas Freiberger5, Sarah E. Leigh6, C. Lisa Kurtz7, Joep Defesche8, Eric J. Sijbrands9, Robert A. Hegele2, Joshua W. Knowles10, Mafalda Bourbon1 on behalf of the ClinGen FH Variant Curation Committee
1Instituto Nacional de Saúde Doutor Ricardo Jorge, Lisbon, Portugal; 2Western University, London ON, Canada; 3Harvard University, Boston MA, USA; 4Hopital Pitie-Salpetriere, Paris, France; 5 Centre for Cardiovascular Surgery and Transplantation,, Brno, Czech Republic; 6Genomics England, London, UK; 7University of North Carolina, Chapel Hill, USA; 8University of Amsterdam, Netherlands; 9University Medical Center, Erasmus MC, Rotterdam, Netherlands; 10Stanford University, Stanford CA, USA.
Familial hypercholesterolemia (FH) is an autosomal dominant disorder of lipid metabolism
characterized by extremely elevated levels of LDL-C and increased cardiovascular risk. A
vast number of potentially pathogenic variants have been identified in FH patients in LDLR,
APOB, and PCSK9 genes. We sought to encourage FH researchers/clinicians worldwide to
submit their variant findings to the centralized ClinVar database, with the ultimate goal of
achieving accurate and consistent variant classification through data sharing and eventual
development of FH-specific variant interpretation guidelines.
There are now 6022 separate submissions and 2920 unique FH-associated variants in
ClinVar from 42 submitters in 14 different countries. The average number of submitters per
variant is ~2. As expected, 80% of unique variants are in LDLR (n=2349), 13% in APOB
(n=365) and 7% in PCSK9 (n=206). In all 3 genes missense variants are the most common
(43% LDLR, 63% APOB, 37% PCSK9), followed by frameshift in LDLR (19%), synonymous
in APOB (21%), and intronic variants in PCSK9 (18%). For variants with multiple submitters,
a concordant pathogenicity classification was achieved for 69% of LDLR variants, but only
for 38% of APOB and 36% of PCSK9 variants. Variant classification by ACMG guidelines
were used by 14 different submitters in 2051 submissions, while the remaining submitters
either used their own criteria (n=16, 2880 variants) or did not provide pathogenicity criteria
(n=12, 1091 variants). For variants classified by ACMG guidelines, 61% were considered
pathogenic/likely pathogenic (P/LP) and 5% benign/likely benign (B/LB). In variants
classified with independent or no identified criteria, 57% were considered P/LP and 17%
B/LB.
In conclusion, this study provides an update on the current state of FH-associated variants
detected in patients worldwide, and highlights the importance of data sharing and
standardized use of variant classification guidelines to further improve FH diagnosis.
In Japan, genetic tests have become standard tools in cancer care. These tests are going to
be covered private as well as government health care plans. There is a high demand of
prompt clinical trial matching services based on genomic variations from clinicians and
patients.
Information on clinical trials in Japan is available in several databases in Japanese, in prior
to registering at ICTRP, WHO. In order to provide updated information, we regularly curate
clinical trial information from multiple databases.
To build a clinical trial matching service, one of the biggest problems is an ambiguity of
description of entries in clinical trials. The entries are written in natural language in various
format. Therefore, simple keyword search is not sufficient to acquire comprehensive search
result.
For example, "EGFR active mutation" is insufficient description for machines. A patient who
has "EGFR p.L858R mutation" or "exon 19 deletion" should be introduced for a clinical trial
of which inclusion criteria has "EGFR active mutation". For another example, medicines
have several aliases on names; code names, brand names (they are different between
countries), abbreviations, synonyms. We need to disambiguate these names.
We developed pipelines to transform clinical trials information to machine-readable
knowledge base based on manually collected information by PhD biomedical scientists.
Using the knowledge base, we constructed automated clinical trials matching services for
cancer patients based on patient genomic variations appearing as inclusion criteria, title etc.,
in clinical trials.
We will present how our knowledge base has developed through manual curation as well as
machine learning technology.
P16
Tools for managing VMC computed identifiers for variant representation
Shawn Rynearson1, Michael Watkins2, Alex Henrie2, Karen Eilbeck2
1. Department of Human Genetics, University of Utah, Salt Lake City, Utah. 2. Department of Biomedical Informatics, University of Utah, Salt Lake City, Utah
The Variant Modeling Consortium (VMC) is a GA4GH guided endeavor to solidify a shared
view of variant representation that will enable reliable data exchange. The specific goal of
the VMC is to define how systems exchange alleles, haplotypes, and genotypes by defining
relevant terms and creating a simple data model using those terms. This group has
developed a specification to standardize the exchange of variation data, contributing agreed
upon terminology, an information model, a machine-readable schema definition and
importantly globally unique computed identifiers. These computed identifiers generated from
object data provide a mechanism to unambiguously name variants. This is a key innovation
as two groups implementing the same algorithms and using the same reference data and
variant objects will generate the same identifiers. This will enable rapid lookups and ease
data sharing. The VMC digest utilizes the SHA-512 hash algorithm to generate a URL
Base64 encoded binary digest, which results in unique IDs for each element of a VMC
object.
We have developed a suite of tools to produce and validate VMC identifiers according to the
VMC specification, to enable rapid development and deployment of VMC enabled software.
The tools provide the following functionality:
VMC Allele IDs for VCF variants. VMC compliant identifiers are added to the INFO field of
each variant in a VCF file, with accompanying VCF Header information.
VMC Sequence IDs for reference sequences. Transform and store VMC hash IDs from
any given fasta record.
Region specific bundle. For a region of VCF, a VMC JSON bundle will be generated.
HGVS conversion. For a simple HGVS expression, a VMC JSON bundle will be
generated.
Validation of user generated ID. Given a user generated ID, validation is provided based
on user provided information and sequence data.
These tools are written in Go and Python and are available from http://vcfclin.org and for
download https://github.com/eilbecklab/VMC-Software-Suite. We anticipate these tools will
accelerate the acceptance of the VMC specification by the bioinformatics and clinical variant
communities and enable the use of computable identifiers across diverse groups for tasks
relating to variant representation and sharing.
P17
Developing a communication rubric for genetic testing - patient-facing curation.
W. Andrew Faucett, Miranda LG Hallquist1, Eric Tricou1, Kyle Brothers2, Curtis R Coughlin II3, Laura Hercher4, Louanne Hudgins5, Howard Levy6, Holly Peay7, Myra Roche8, Melissa Stosic9, Maureen Smith10, Wendy Uhlmann11, Karen Wain1, Kelly E Ormond5, Adam H Buchanan1
1Geisinger Health System, 2University of Louisville, 3University of Colorado, 4Sarah Lawrence College, 5Stanford University, 6Johns Hopkins University, 7RTI International, 8University of North Carolina, 9Columbia University., 10Northwestern University, 11University of Michigan
As genes are curated by ClinGen and other groups, developing care delivery models that
increase access to genetic testing and genetic counseling is critical for expanding the role of
genetics in routine healthcare. ClinGen's Consent and Disclosure Recommendations
working group (CADRe) has developed and evaluated rubrics for determining a suggested
communication approach for genetic testing consent and disclosure. We propose three
possible communication levels: (1) traditional genetic counseling (TGC) with a genetics
specialist, (2) targeted discussion with an ordering clinician, or (3) brief communication
supported by educational resources. The CADRe recommendations provide guidance
regarding which genetic conditions and testing indications would benefit most from TGC
(where detailed discussion, complicated test selection, and psychosocial support are
provided), with the goal of directing genomics expertise to those patients for whom it is most
impactful. The CADRe workgroup has tested the application of the model by reviewing the
ACMG Secondary Findings v2.0 gene list, examining each gene in the context of specific
indications for genetic testing, including: confirmation of a clinical diagnosis, testing an
individual with a suggestive personal history, testing an unaffected individual with a
suggestive family history, and testing an unaffected individual for a known familial variant.
Results of this exercise among these medically actionable genes suggest that much of the
pre-test genetic counseling can be triaged and transitioned to targeted discussions with
ordering healthcare providers. Additionally, the indication for testing appears to play a more
significant role in the model than initially anticipated. Determining the level of communication
between the healthcare provider and the patient considering genetic testing is important to
provide patients with adequate education and support. Continued curation of the
communication approach for specific genetic tests may shift the paradigm of genetic testing
to emphasize the use of genetics providers in complex cases that require specialized
genetics expertise including the return of pathogenic results.
P18
What's in a name? OMIM's approach to naming Mendelian phenotypes
Ada Hamosh, Carol Bocchini, Anne Stumpf, Marla O'Neill, Cassandra Arnold, Joanna Amberger
Institute of Genetic Medicine, Johns Hopkins University, Baltimore, Maryland, USA
OMIM, a catalog of human genetic phenotypes and their underlying genes, has been
engaged in the process of naming genetic phenotypes for over 50 years. This process is a
complex undertaking with many stakeholders. While it is tempting to include clinical,
pathologic, and molecular information within a single name, it is not practical or advised, not
least because the knowledge of any particular phenotype changes over time. Ideally, a name
should be mnemonic, conjuring up an image of the phenotype, and euphonious. Phenotype
names in OMIM represent part of the catalog's organizing principles, and new names are
given within the context of existing phenotypes. When considering a new phenotypic entity
for a particular gene, OMIM first reviews the range of disorders already known to be
associated with the gene. A phenotype must be clearly distinct from those already described
to justify creation of a new phenotype. Highly related genetically heterogeneous phenotypes
are organized into numbered Phenotypic Series; no hierarchy in these series is implied. If a
similar disorder does not exist in OMIM, a name will be given to it using one of the following
methods: 1) a listing of cardinal clinical features, preferably leading to a memorable
acronym; 2) eponyms based on the authors who first described the condition, thus tying the
publication to the phenotype; 3) when feasible, the naming conventions defined by
subspecialty groups and experts; and 4) in some cases, particularly enzyme deficiencies,
based on the basic defect. OMIM avoids naming disorders after the mutated gene. Naming
genes, the responsibility of the HGNC, is itself complex, and names are appropriately
changed over time as more is learned about a gene's function(s). The same designation for
a gene and a phenotype obfuscates molecular and medical concepts and promotes
confusion in the literature, particularly because more than one-third of known disease genes
cause more than one phenotype, each with unique features, prognoses, and treatment.
Whereas some medical terms and published designations may be objectionable to patients,
we make efforts to revise or avoid these while maintaining some naming stability. Evolving
clinical designations are retained as alternative titles under a stable MIM number.
P19
Improving classification of truncating variants in autosomal dominant hearing loss genes using patient and population variant data
Sarah E. Hemphill, Andrea M. Oza, Marina T. DiStefano, Ahmad N. Abou Tayoun, Heidi L. Rehm and Sami S. Amr
1Laboratory for Molecular Medicine, Partners Healthcare Personalized Medicine, Cambridge, MA, USA. 2Division of Genomic Diagnostics, The Children's Hospital of Philadelphia, The University of Pennsylvania Perelman School of Medicine, Philadelphia, PA, USA. 3Department of Pathology, Brigham & Women’s Hospital, Harvard Medical School, Boston, MA, USA 4Center for Genomic Medicine, Massachusetts General Hospital, Boston, MA, USA 5The Broad Institute of MIT and Harvard, Cambridge, MA, USA
Autosomal dominant hearing loss (ADHL) accounts for an estimated 20% of genetic hearing
loss. Over 40 genes have been associated with syndromic and/or nonsyndromic ADHL with
many exhibiting variable onset and expressivity, and some are also associated with
autosomal recessive hearing loss (ARHL). The molecular mechanism of disease for most of
these genes is unclear. Thus, the clinical significance of truncating variants in ADHL genes
is often uncertain, particularly in genes which cause both ADHL and ARHL.
To address this issue, we developed a framework to determine whether haploinsufficiency
caused by loss-of-function (LoF) sequence variants is disease-causing based on the
location, number, and type (nonsense, frameshift, splice) of pathogenic LoF variants in the
literature and applied it to 39 ADHL genes. In addition, we assessed the utility of LoF
intolerance scores (pLI) from the Exome Aggregation Consortium (ExAC) in genes that we
determined to cause ADHL by haploinsufficiency. Since the pLI score uses counts of rare
variants and is optimized for severe disease, we instead calculated a total LoF minor allele
frequency (LoFMAF) for each gene by summing the MAFs of all LoF variants in all
populations in the Genome Aggregation Database (gnomAD).
Of 39 ADHL genes, 10 genes met our criteria for haploinsufficiency as a disease mechanism
(ADHL haploinsufficiency genes) based on evidence from the literature. In 11 genes, LoF
caused ARHL only, 3 genes caused ADHL by another mechanism, and the remaining 15
genes did not have sufficient evidence in the literature to support nor refute
haploinsufficiency.
Comparing the results of our literature curation to the ExAC pLI and gnomAD dataset
revealed that pLI scores had low sensitivity (70%, 95%CI: 35-93%) but high specificity (95%,
95%CI: 86-99%) in predicting ADHL haploinsufficiency genes. Finally, analysis of the total
LoFMAFs in ADHL genes showed significantly lower LoFMAFs in ADHL haploinsufficiency
genes compared to LoFMAFs in non-haploinsufficiency ADHL and/or ARHL genes (unpaired
t-test, p=0.03), with MYO6 having the highest total LoFMAF of ADHL haploinsufficiency
genes at 0.037%.
In summary, this framework provides a strategy for determining whether novel LoF variants
are likely to cause ADHL when functional studies clarifying disease mechanism are
unavailable. In addition, we show that the total LoFMAFs for ADHL haploinsufficiency genes
are significantly lower than other HL genes, and that this metric could be used to set a
threshold to rule out haploinsufficiency for novel ADHL genes with an unknown disease
mechanism.
P20
Pathogenicity interpretation for two de novo mutations in Caudal Type Homeo Box transcription Factor 2 (CDX2) in patients with persistent cloaca. Jacob Shujui Hsu1,2,6, Manting So3, Clara S. M. Tang3,6, Anwarul KARIM3, Robert Milan Porsch2,6, Carol Wong3, Michelle Yu3, Fanny Yeung3, Hui-min Xia4, Ruizhong Zhang4, Stacey Shawn Cherny2,6, Patrick Ho Yu Chung3, Kenneth K. Y. Wong3, Pak C. Sham2,6, Ngoc Diem Ngo5, Miaoxin Li2,6, Paul K. H. Tam3,6, Vincent C. H. Lui3, Maria-Mercè Garcia-Barcelo3,6
1Departments of Medical Genetics and Internal Medicine, National Taiwan University
Hospital, Taipei, Taiwan; 2Department of Psychiatry, Li Ka Shing Faculty of Medicine, The
University of Hong Kong, Hong Kong, China; 3Department of Surgery, Li Ka Shing Faculty of
Medicine, The University of Hong Kong, Hong Kong, China; 4Guangzhou Women and
Children's Medical Center, Guangzhou, Guandong, China; 5National Hospital of Pediatrics,
Ha Noi, Viet Nam; 6Centre for Genomic Sciences, Li Ka Shing Faculty of Medicine, The
University of Hong Kong, Hong Kong, China
The cloaca is an embryonic cavity that is divided into the urogenital sinus, vagina, and
rectum upon differentiation of the cloacal epithelium triggered by tissue-specific transcription
factors including CDX2. Defective septation anomalies during the development lead to
persistent cloaca (PC) in humans. Although no gene has ever been identified, there is a
strong evidence for a genetic contribution to PC from mouse models. We applied whole-
exome sequencing and copy-number-variants analyses to 20 PC patients and their
unaffected parents. The novel p.Cys132* and p.Arg237His de novo CDX2 variants were
identified in two unrelated patients. Both variants were novel among hundred thousand
whole exome sequencing sample in the gnomAD database (N >123,136). Not only the
predictive model from Phyre2 Investigator indicated the p.Arg237His is located on DNA
binding domain with higher mutation sensitivity compared with adjacent regions, but also
SWISS-MODEL predictive protein secondary structure indicated the presence of a structural
change in the DNA binding domain. Moreover, both variants altered the expression of
CYP26A1, a direct CDX2 target encoding the major retinoic acid (RA)-degrading enzyme. In
spite of the fact that gene constraint score indicated normal gene-level pathogenicity for
CDX2 gene, which implied narrow pathogenicity effect by denovo mutation, another machine
dominant and pediatric pathogenicity. Other genes governing the development of cloaca-
derived structures were recurrently mutated and over-represented in the extracellular matrix-
receptor interaction pathway (MsigDBID: M7098, FDR: q-value < 7.16 × 10-9). Given the
CDX2 de novo variants and the role of RA, our observations could potentiate preventive
measures. This is the first evidence that PC is genetic, with genes involved in the RA
metabolism at the lead. For the first time, a gene recapitulating PC in mouse models is found
mutated in humans. On the other hand, despite the severity, rarity and heterogeneity of PC,
establishing disease causality for any given gene is extremely difficult. It might be required to
screen a large number of patients with identical phenotype for achieving genome-wide
statistical power or to perform sophisticated functional assays to establish the disease
causality for each gene. As patients with the identical rare disease are limited, data sharing
and re-analysis should be fully considered and should be conducted under proper
regulations.
P21
Integrative analysis of cancer genome profiling data to study the interplay of genetic background and molecular mechanisms in cancer
Qingyao Huang, Michael Baudis
Institute of molecular Life Sciences, University of Zürich, Zürich, CH-8057, Switzerland; Swiss Institute of Bioinformatics, University of Zürich, Zürich, CH-8057, Switzerland
Genetic mutations accumulate during the formation of malignant neoplasia. Endogenous
processes, such as inflammation and exogenous agents, such as chemical carcinogens or
UV radiation, accelerates DNA damage and causes genome modification. Thanks to the
recent efforts in sequencing the cancer genomes, researchers have started to comprehend
patterns of point mutation in various cancer types. However, the copy number variation
(CNV) patterns, which comprise a large part of these variations, is less studied.
Those novel mutations emerging during one's lifetime, termed "somatic" mutations, can be
influenced by inherited ("germline") genome variations. The germline variations are
determined by the ethnic background of individuals, thereby associated with their
geographical location. Although socio-economic components contribute to disease incidence
and mortality in general, several inherited single nucleotide variation (SNV) are found highly
associated with developing specific types of cancer, motivating a more thorough search in
germline roots for somatic variation patterns.
With a combination of ~50,000 curated oncogenomic array data from the arrayMap database
and ~20,000 profiles from TCGA project depository, we perform a meta- analysis to
investigate influence of genetic background on the CNV patterns in cancer. From
sequencing data of 26 world-wide populations from 1000 Genomes project, we extract the
SNP markers and use them for subsequent sample analysis. First, we show that using
admixture analysis, the population classification is accurate even from low- resolution arrays
(10k markers). This appends genome-derived population information to the database, as an
additional layer to the geographic location of each sample. Next, we link various types of
CNV to the identified population group to discover potential population-specific oncogenic
patterns. We utilize a deep learning approach, i.e. autoencoder, to reduce noise and
complexity in the CNV pattern and extract the abstract features in the CNV pattern. We also
look into individual cancer types based on their NCIT classification to explore hints and
significance of population background.
P22
ClinGen’s Pediatric Actionability Working Group
Jessica Hunter, Elizabeth Webber1, Kathleen Mittendorf1, Kristy Lee2, Marc Williams3, Bradford Powell2, Katrina Goddard1
1 Center for Health Research, Kaiser Permanente Northwest, Portland, OR, USA; 2 Department of Genetics, University of North Carolina, Chapel Hill, NC, USA; 3 Genomic Medicine Institute, Geisinger, Danville, PA, USA
ClinGen established the Pediatric Actionability Working Group (PAWG) to assess the clinical
actionability of secondary genomic findings in children and adolescents. The workgroup will
adapt the existing adulthood-focused framework of the Actionability Working Group (AWG)
while accounting for factors specific to actionability in the pediatric population, such as the
principle of maintaining an open future related to genetic conditions not actionable until
adulthood. The AWG framework uses standardized methods to curate evidence for four
domains of actionability: 1) severity of the outcome; 2) likelihood of the outcome
(penetrance); 3) effectiveness of the intervention to prevent harm; and 4) nature of the
intervention (risk/burden to the individual). A semi-quantitative metric is applied to generate
consensus scores for each domain. The PAWG will curate evidence and score actionability
related to implementation of clinical interventions during the pediatric period that lead to
disease prevention or delayed onset and improve downstream clinical outcomes for genetic
disorders. This scope includes genetic disorders with outcomes with pediatric onset. This
scope also includes disorders with outcomes which typically do not present until adulthood if
there is evidence that an intervention during childhood or adolescence can optimize
outcomes (e.g., use of statins in familial hypercholesterolemia). Accordingly, we will consider
lifetime penetrance, rather than age-related penetrance, when scoring likelihood of the
outcome. The PAWG will focus on assessing actionability of interventions during childhood
or adolescence that relate to patient management, surveillance, and circumstances to avoid.
While the original AWG protocol included recommendations related to family management
(e.g., genetic testing of at-risk adult relatives), these recommendations are not actionable in
pediatric patients themselves. Thus, recommendations related to family management or
recommendations deferred until adulthood will be excluded from PAWG assessments. The
scope of the PAWG protocol targets secondary findings in pediatric patients undergoing
clinically indicated diagnostic testing. Importantly, it does not capture all factors relevant to
population-based screening (e.g., newborn screening), and is not a sufficient determinant for
recommending screening in asymptomatic cohorts. The curation provided by the PAWG will
support research and clinical communities in making decisions and recommendations about
reporting secondary findings from genome-scale sequencing in pediatric populations.
P23
Whole exome sequencing of thirty adults with different patterns of adult onset hearing loss
Morag A. Lewis (1,2), Lisa S. Nolan (3), Barbara Cadge (3), Lois J. Matthews (4), Bradley A. Schulte (4), Judy R. Dubno (4), Karen P. Steel (1,2), Sally Dawson (3)
1. Wolfson Centre for Age-Related Diseases, King`s College London, SE1 1UL, UK 2. Wellcome Trust Sanger Institute, Hinxton, Cambridge, CB10 1SA, UK 3. UCL Ear Institute, University College London, WC1X 8EE, UK 4. The Medical University of South Carolina, SC, USA
Hearing loss is one of the most common sensory deficits in the human population, and it has
a strong genetic component. However, although to date more than 140 loci relating to
human hearing loss have been mapped, and over 100 genes identified, the vast majority of
genes involved in hearing remain unknown. In order to explore the landscape of variation
associated with hearing loss, we sequenced the exomes of thirty patients selected for
distinct phenotypic sub-types from well-characterised cohorts of 1479 people with adult-
onset hearing loss. After sequencing, variants were called with SAMtools and Dindel, and
filtered based on quality, frequency in the non-Finnish European population, predicted
consequence and predicted severity of impact. We examined the results for genes which
carried mutations in more than one individual and also compared them to a list of genes
known to be associated with deafness in mice or humans.
From these comparisons, we have identified multiple candidate mutations for further
investigation and follow-up. We also found that every patient carried predicted pathogenic
mutations in at least ten deafness-associated genes; similar findings were obtained from an
analysis of the 1000 Genomes Project data unselected for hearing status. The high
frequency of predicted-pathogenic mutations in known deafness-associated genes in the
population was unexpected and has significant implications for current diagnostic
sequencing in deafness. Our results illustrate the complexity of genetic contributions to
hearing loss and the power of stratified analysis in complex disease to identify candidate
variants for further study.
This work was supported by the following: NIH/NIDCD P50 000422; the Wellcome Trust
(100669); the Haigh Fellowship in age related deafness, Deafness Research UK.
P24
UniProt: enabling interpretation of protein variation effects
Michele Magrane1, Andrew Nightingale1, Peter McGarvey3, Sandra Orchard1, Maria Martin1, UniProt Consortium1,2,3
1European Bioinformatics Institute (EMBL-EBI), European Molecular Biology Laboratory, Wellcome Trust Genome Campus, Hinxton, Cambridge CB10 1SD, UK 2Swiss Institute of Bioinformatics, Centre Medicale Universitaire, 1 rue Michel Servet, CH-1211 Geneva 4, Switzerland 3Protein Information Resource, Georgetown University Medical Center, 3300 Whitehaven St. NW, Suite 1200, Washington, DC 20007, USA and University of Delaware, 15 Innovation Way, Suite 205, Newark, DE 19711, USA
Understanding the effect of genetic variants on protein function is crucial to a thorough
understanding of the role of proteins in disease biology. UniProt provides the scientific
community with a comprehensive, high-quality and freely accessible resource of protein
sequence and functional information. It aims to support clinical researchers by providing a
wealth of variation data coupled with information about how these variants affect protein
function. A team of expert biologists reviews and compiles published variants and their
functional effects from the scientific literature. This is combined with large-scale variation
data imported from a variety of sources including the 1000 Genomes project, COSMIC, the
Exome Aggregation Consortium (ExAC), the Exome Sequencing Project (ESP) and ClinVar
to provide a comprehensive catalogue of variation data which is freely available from the
UniProt website at www.uniprot.org. To facilitate interpretation of variant data, UniProt
provides a number of tools to allow visualisation of variants in the context of other protein
information and to integrate UniProt data into external tools and workflows. Genome tracks
are provided to allow integration of UniProt data into genome browsers such as those
provided by Ensembl and UCSC. The UniProt protein sequence viewer, ProtVista, provides
a graphical visualisation of protein sequence features from multiple sources in a single view
and is made freely available so that users can add their own data and integrate it into other
web resources. Programmatic access is provided by the UniProt Proteins API
(http://www.ebi.ac.uk/proteins/api), a REST interface which allows users with little or no
programming background to integrate a broad range of biological data into their analyses.
Future developments include a Protein Variant Effect Predictor which will integrate genome,
protein and structure data to enhance interpretation of variant effects. Through provision of
extensive variant data and user-friendly tools, UniProt supports clinical researchers by
enhancing understanding of the link between variation and protein function.
P25
PanelApp: A Community-Curated resource for the Scientific and Clinical Community for Genome Analysis, Interpretation and Actionability
Ellen McDonagh, Ellen M. McDonagh1,2,, Antonio Rueda1, Helen Brittain1, Louise C. Daugherty1,2, Rebecca E. Foulger1,2, Kristina Garikano1, Oleg Gerasimenko1, Sarah Leigh1, Olivia Niblock1, Richard H Scott1, Damian Smedley1,2, Ellen R A Thomas1, Arianna Tucci1, Eleanor Williams1,2, Mark J Caulfield1, Augusto Rendon1,2
Genomics England, 1Queen Mary University London, Dawson Hall, London, UK 2The Biodata Innovation Centre, Wellcome Genome Campus, Cambridge, UK.
Genomics England PanelApp (https://panelapp.genomicsengland.co.uk/) is a unique open
source Knowledgebase that enables crowdsourcing of evidence-based review from global
Clinicians and Researchers to create diagnostic-grade virtual gene panels for diseases. It
currently has 213 gene panels covering over 2300 OMIM diseases with a total of 4186
genes, more than 800 registered reviewers and 12,853 external reviews. PanelApp is
integrated into the Genomics England genome analysis pipeline, aiding variant prioritisation
to provide interpretation results to clinicians within the National Healthcare System (NHS) as
part of the 100,000 Genomes Project and will help support the provision of commissioned
genomes by NHS England's planned Genomic Medicine Service.
We are now developing PanelApp to move beyond genes, to curate genomic regions of
clinical importance such as STRs, enhancing the scope and value of the PanelApp resource
as well as the Genomics England genome interpretation pipeline. Genes linked to known
interventions or therapies are also curated, and future developments will expand on this to
include clinically actionable information within the genome such as pharmacogenetic panels
and links to gene therapy trials. Integration of input from the Genomics England analysis
pipeline and from patient diagnoses are fed back into PanelApp to enhance the
interpretation analysis pipeline and complete the knowledge feedback loop. In addition, as it
is open source, NHS Bioinformaticians are utilising PanelApp for analyses of their own omics
data, Clinical Interpretation Partners have integrated the gene panels into their systems, and
PanelApp contributes to well established international databases such as Open Targets
(contributing to drug discovery) and DECIPHER (aiding clinical decision making for
diagnoses). In the near future, novel genes and discoveries published by the research
endeavors of the Genomics England Clinical Interpretation Partners (GeCIPs) generated
from the 100,000 Genomes data will be incorporated.
Our dynamic and continual internal scientific curation of evidence-level assessment collates
knowledge on gene-disease relationships from the scientific literature, other key curated
resources, and external expertise to allow rapid update of which genes have enough
evidence for clinical reporting.
P26
Chances and challenges of high-throughput sequencing of Mendelian disorders
Janine Meienberg1, Anna M. Kopps1, Michel Plüss1,2, Sylvan M. Caspar1, Nicolo Dubacher1, Gabor Matyas1,3
1Center for Cardiovascular Genetics and Gene Diagnostics, Foundation for People with Rare Diseases, Schlieren-Zurich, Switzerland; 2Institute of 4D Technologies, University of Applied Sciences and Arts Northwestern Switzerland, Windisch, Switzerland; 3Zurich Center for Integrative Human Physiology, University of Zurich, Zurich, Switzerland
High-throughput sequencing (HTS) is widely used for clinical applications such as the
molecular diagnosis of Mendelian disorders. As the applied technology/workflow
substantially affects the diagnostic yield, knowledge about the pitfalls and advantages of
HTS technologies and analysis pipelines is crucial for the successful application of hitherto
unprecedented large-scale genetic testing.
We address the chances and challenges of HTS in the molecular diagnosis of Mendelian
disorders as well as assess the sensitivity/recall, precision, computation time, and disk
footprint of four corresponding HTS analysis pipelines.
We exemplify the limitations of targeted (gene panel) and whole-exome sequencing (WES)
as well as emphasize the potential of whole-genome sequencing (WGS) in the detection of
single nucleotide variants (SNVs) and copy number variations (CNVs). In addition, we
elucidate limitations of short-read HTS on exemplary cases including the influence of
homologous/repetitive regions (mappability <1) on variant calling and the impact of
sequence composition on read depth, as well as show differences in the performance of
WGS analysis pipelines.
We recommend to select the HTS method with care and to combine more than one
independent bioinformatics pipeline for the most comprehensive data analysis. The use of
PCR-free WGS (>60×) instead of WES or panels and the inclusion of CNV analysis can
contribute to increased diagnostic yield in molecular diagnosis with lifetime value. As long-
read HTS may overcome limitations of short-read HTS, it is envisioned as the future of
(clinical) sequencing.
P27
Building whole genome sequencing capacity in Scottish research and healthcare
Alison M Meynert, The Scottish Genomes Partnership Consortium
MRC Human Genetics Unit, MRC Institute of Genetics and Molecular Medicine, University of Edinburgh, Edinburgh, UK
The Scottish Genomes Partnership (SGP) is a major Scotland-wide research programme
between Scottish universities and NHS Scotland (NHSS) to sequence human genomes for
research and to build clinical whole genome sequencing capacity. The SGP is funding the
whole genome sequencing of population isolates (Viking study of Shetland islanders), rare
diseases in a research context (eye malformations, motor neuron disease, microencephaly,
sex differentiation), cancer (pancreatic, oesophageal, and ovarian) and clinical patient
sequencing in the NHSS.
Samples are sequenced at either Edinburgh Genomics or the Glasgow Precision Oncology
Laboratory and delivered as either raw reads (FASTQ files) or aligned reads (BAM files) with
variant calls (VCF files) from a standard best practice pipeline. SGP funded bioinformatics
teams then carry forward custom analysis for each project: joint genotyping and annotation
for population isolates, familial joint genotyping and inheritance model filtering for rare
diseases and clinical patients, and somatic structural and small variant calling for cancer
samples.
The Edinburgh arm of the partnership is developing a securely hosted Scottish variant
repository to warehouse and provide the germline variants from appropriately consented
projects for use within the SGP and longer term. Variants from other Scottish sequencing
projects, for example the Lothian Birth Cohort (www.lothianbirthcohort.ed.ac.uk), will also be
stored in the repository. The variant repository will initially provide aggregate allele
frequencies for each input cohort via an instance of the OpenCB Interactive Variant Analysis
browser (github.com/opencb/iva). We plan to further develop the resource so that
researchers can securely interrogate variants from their projects on an individual and family
level.
SGP is funded by the Chief Scientist Office of the Scottish Government Health Directorates
[SGP/1] and The Medical Research Council Whole Genome Sequencing for Health and
Wealth Initiative.
P28
Identification of non-deletional Thalassaemia mutations by next-generation sequencing
Multiple candidate variants from whole genome sequencing analysis in a family with autism spectrum disorders
Dr Ying Qiao, Ying Qiao1, Ryan KC Yuen2, Robert M. Stowe3, Kristina Calli1, Sally Martell1, Chieko Chijiwa1, Evica Rajcan-Separovic4, Stephen W Scherer2, Suzanne Lewis1
1Medical Genetics, University of British Columbia (UBC), Vancouver, BC, Canada; 2The Centre for Applied Genomics, The Hospital for Sick Children, Toronto, Ontario, Canada; 3Psychiatry and Medicine (Neurology), UBC, Vancouver, BC, Canada; 4Pathology and Laboratory Medicine, UBC, Vancouver, BC, Canada
Whole genome sequencing (WGS) has been more widely used as a tool in the clinical
diagnosis and it helps increase the diagnosis rate up to 20% in autism spectrum disorders
(ASDs). Thus far, more than 100 genes and CNV loci have been reported to be associated
with ASD suscpeptibility. However, none are found in >1% of cases with ASDs, suggesting a
diverse genetic heterogeneity of the disorders. We have recruited >700 families with simplex
and multiplex ASD and are performing chromosome microarray and whole genome
sequencing studies as part of the iTARGET Autism project (http://www.itargetautism.ca/) and
the MSSNG project (https://www.mss.ng/), an open-access genomic database for autism
research. Using WGS in a trio ASD family in combination with internal bioinformatics
pipelines and a commercial software VarSeq, we identified 4 inherited rare damaging
missense single nucleotide variants (SNVs) in 4 genes (DNMT3A, PHF2, NRXN2, and
SNRPN), and one rare copy number variant (CNV). The CNV is a paternally inherited 14 Kb
microdeletion in ZNF517, which is also confirmed by DNA microarray.The variants in
DNMT3A and SNRPN were also confirmed by Sanger sequencing. Clinically, the male
proband has ASD, moderate intellectual disability, developmental delay, verbal apraxia,
post-natal macrocephaly, large stature, adult-onset epilepsy (age 22 years), and mild facial
dysmorphism (round facies, bitemporal narrowing, narrow palpebral fissures, low-set and
protuberant ears, hypotonia, high arched palate). Neither parent has intellectual disability or
ASD. All of the genes involved in these rare variants and CNV are reported to be ASD-
related and involved in brain/neuron development. De novo mutations in the above SNV
genes and a recurrent deletion in ZNF517 gene have been identified in cases with ASDs. In
our proband, mutations in DNMT3A, PHF2, and SNRPN are paternally inherited while
NRXN2 is maternal. DNMT3A is a newly identified ASD candidate gene and its mutation is
associated with Tatton-Brown-Rahman Syndrome. Some of the phenotypes are shared in
our proband. Functional analysis is in progress including whole transcriptome analyses.
Conclusion: DNMT3A is likely the most relevant gene accounting for both ASD and features
concordant with Tattan-Brown-Rahman syndrome. Alternatively, our subject's ASD
phenotype reflects a collection of quantitative phenotypic traits associated with each of the
multiple ASD risk genes identified and its complex genetic origins.
P31
Somatic Variant Data Integration in ACMG Classification of Germline Variants in Cancer Susceptibility Genes
Dr. Deborah I. Ritter, Deborah I. Ritter1, Chimene Kesserwan2, Dmitriy Sonkin3, Debyani Chakravarity4, Elizabeth Chao5, Raj Ghosh6, Kristy Lee7, Shashi Kulkarni6, Liying Zhang4, Kenneth Offit4, Sharon E. Plon1, Michael F. Walsh4
1Texas Children's Hospital and Baylor College of Medicine, USA; 2St. Jude's Children's Research Hospital, USA; 3National Cancer Institute, USA; 4Memorial Sloan Kettering Cancer Center, USA; 5University of California, Irvine, USA; 6Baylor College of Medicine, USA; 7University of North Carolina, USA
Curation of germline variants in cancer susceptibility genes is critically important for
identifying underlying cancer predisposition syndromes and may alter clinical management.
Somatic mutation data may substantially inform germline interpretation, but currently lacks
standardized use. In particular, the American College of Medical Genetics and Genomics
(ACMG) and Association of Molecular Pathology (ACMG-AMP) variant interpretation
guidelines do not incorporate somatic (tumor) data in germline variant interpretation. The
Clinical Genome Resource Germline/Somatic Variant Curation subcommittee (GSVC) has
undertaken a dedicated effort to provide guidance on the integration of somatic data in
hereditary cancer variant interpretation, and understand principal usage caveats of somatic
data. The GSVC includes 18 members across 12 institutions with expertise spanning
pathology, laboratory diagnostics, bioinformatics, medical genetics and oncology. We
circulated a somatic data usage survey to professionals involved in germline variant
interpretation at cancer centers. Of 21 respondents, 16 (76.9%) reported following ACMG
guidelines and 13/16(81%) reported interest in incorporating somatic data for germline
variant classification. Additional questions were posed regarding types of somatic data, and
responses guided our review of somatic data features. The GSVC then conducted an
interpretation exercise on ~45 variants across oncogenes and tumor suppressors to explore
somatic data elements for germline interpretation. By comparison of mutational data in
cancerhotspots.org, germline functional assays and ClinVar interpretations we defined an
optimized use of somatic hotspot data when interpreting germline variants in hereditary
cancers. We propose a conservative approach limited to using the existing PM1 and
PM1_Supporting evidence codes, and provide guidance on optimal use of somatic hotspot
data. We reviewed the many parameters associated with use of loss of heterozygosity (LOH)
data, but at the present time we do not propose a standardized LOH incorporation. By
careful consideration of somatic data elements, and curation testing to understand
incorporation, we aim to ensure maximized and standardized use of available somatic data
for the interpretation of variants in hereditary cancers.
P32
Genetic Dichotomy
Helen Savage
Deputy Head of Clinical Services, Congenica, UK
A self-taught geneticist diagnosed herself and her three siblings with an ultra-rare form of
EDMD symptoms, e.g. partial lipodystrophy, the siblings displayed some drastically different
phenotypes. In particular two siblings displayed muscular wasting, with one wheelchair
bound from the age of 33, whilst the others displayed hyper-muscularity.
To confirm her self-diagnosis, the woman, Jill Viles, sent her and her family’s samples to
Istituto di Genetica Biochimica ed Evoluzionistica, in Bologna, Italy. Results confirmed her
diagnosis of EDMD, and showed that the three siblings had the same missense mutation in
the LMNA gene. The mutation explained their EDMD diagnosis, but not their differing
phenotypes, but with no further funding, Jill’s investigations came to an end.
Hearing of Jill’s self-diagnosis story, a UK-based company, Congenica, reached out to assist
in the investigation of the underlying cause of the differences in phenotypes. To do so they
used, Sapientia, a world-leading clinical decision support platform for interrogation and
analysis of rare inherited disease. The Clinical Scientists at Congenica investigated the
phenotypic differences by first looking at genes associated with neuromuscular conditions
including EDMD, lipodystrophy, myopathy and other neuromuscular phenotypes. They did
not find any variation that would lead to such a difference in phenotypes, or evidence of a
second disorder segregating in the family. They next considered genes known to be
associated with muscle development, and then extended the analysis to cover other genes
acting in the same biological pathways, searching for a potential modifier gene.
Using their many years of experience and a comprehensive set of filters in Sapientia, the
team found a single missense variant (Q311R) in the SMAD7 gene that was present in the
two siblings with muscular wasting, but absent in those with hypermuscularity. The SMAD7
gene, which forms part of the TGFBeta pathway, is involved in skeletal muscle growth and
development. The variant had not been previously reported in the literature, was absent from
gnomAD and affected a well-conserved amino acid; the residue is conserved to zebrafish.
The gene has a significant ExAC missense constraint score of 3.87, indicating missense
changes may be associated with a deleterious effect on the protein.
SMAD7 competes with TGFBeta and myostatin signalling by competing with R-smads for
binding with the type 1 receptor. Other studies have suggested that SMAD7 enhances
skeletal muscle differentiation and is required for the formation of muscular tissue 1,2,3.
With this body of evidence, the team agreed that the result was pertinent enough to report
back to the family as the potential cause for the differing phenotypes. Jill is pursuing this
finding, which is now forming the basis for new Drosophila and cohort studies led by Dr Lori
Walrath and Dr. Benjamin Darbro of the University of Iowa.
1. Cohen et al 2015 Genetic disruption of Smad7 impairs skeletal muscle growth and regeneration J Physiol. 593(Pt
11): 2479–2497.
2. Hua et al 2016. SMAD7, an antagonist of TGF-beta signalling, is a candidate of prenatal skeletal muscle
development and weaning weight in pigs. Mol Biol Rep. 43(4):241-51
3. Winbanks et al 2016. Smad7 gene delivery prevents muscle wasting associated with cancer cachexia in mice. Sci
Transl Med 8(348): 348ra98
P33
Swiss Variant Interpretation Platform (SVIP)
Daniel J. Stekhoven, Patrick Ruch, Valérie Barbié
NEXUS Personalized Health Technologies, ETH Zurich, Zurich, Switzerland; SIB Swiss Institute of Bioinformatics, Switzerland; HES-SO/HEG, Geneva, Switzerland
In 2015, the Clinical Bioinformatics group of the Swiss Institute of Bioinformatics (SIB)
launched a working group for somatic mutation calling, in order to harmonize and improve
NGS practices and foster a community in oncology and hemato-oncology across Swiss
hospitals. The group comprises medical and computational experts from all universities and
many other medical institutions. A key insight from their work was the absence of a central
repository of clinically verified variants in patients. Subsequently, optimal translation of NGS
results into medical practice was hindered.
The suggestion of centralising somatic variants in one single place, harmonising their
annotation, mutually agreeing on their clinical interpretation, and using SIB resources to
support the curation of previously undescribed variants has been accepted as an
infrastructure development project as part of the national Swiss Personalized Health
Network (SPHN, www.sphn.ch) initiative.
The Swiss Variant Interpretation Platform (SVIP) will provide a joint knowledge base for
somatic variants found in Swiss hospitals during cancer diagnostic sequencing. Submission
of new variants will be batch based and coupled with the retrieval of database contents
capturing annotations and interpretations for the given set of variants. This will be further
enriched with an API enabling seamless integration into existing pathology information
systems. SVIP will incorporate variant information from other similar projects such as
ClinVar, ClinGen, CIViC, OncoKB, and PMKB, to facilitate the prioritization of variants by
molecular pathologists.
In an initial ramp up, SVIP will reconcile all previous somatic variants of the partner hospitals
to provide a harmonized annotation. In addition to increasing the frequency of some rare
variants, this step will make it possible to identify conflicting annotations in partnering
institutions. Discrepancies will then be resolved by a clinical expert panel. The panel will also
validate new annotations recommended by the SVIP curation team. Finally, SVIP will offer a
finely customisable notification framework which can inform medical institutions on changes
in annotation of earlier submissions.
SVIP is an ambitious project to establish a Swiss one-stop shopping for the interpretation of
somatic variants, enabling faster and more robust prioritisation. A high-quality, joint variant
annotation pipeline will ensure reproducibility and consistent data stewardship. The secure
interpretation transaction space for molecular pathologists and oncologists will make it
possible to establish a continuous learning system, contributing to improved interpretation of
variants also globally.
P34
Leveraging existing clinical information systems for semi-automated preparation of ClinVar submissions.
Timothy Tidwell1, Sara Brown1, Genevieve Pont-Kingdon1, Zoe Lewis1, Erica Andersen1, 2, Rong Mao1, 2, Elaine Lyon1, 2
1ARUP Laboratories, Salt Lake City, Utah, USA; 2 Department of Pathology, University of Utah, Salt Lake City, Utah, USA
Data sharing through ClinVar is an important step in the Next Generation Sequencing (NGS)
laboratory process for quality management but is not easily accomplished. Submissions to
ClinVar can be labor intensive, requiring manual curation of spreadsheets containing
variants, classifications, and evidence for or against pathogenicity. However, much of the
information submitted to ClinVar is required and stored by tools that are used to generate
clinical reports. At ARUP Laboratories, we have developed a system to export existing data
into a spreadsheet that can be submitted directly to ClinVar.
Data generated from the bioinformatics pipeline are stored in an internal database and are
accessible through an in-house website (NGS.Web). Variant curation is performed within
NGS.Web, and variant classifications and comments (evidence) are stored in the database.
These data are used to create a clinical report, as well as, to generate a ClinVar submission.
By streamlining report generation and variant submission processes, this system reduces
both time spent generating ClinVar entries, as well as, the potential for data entry errors, as
all fields are thoroughly reviewed in clinical reporting. Through this semi-automated process,
the frequency of ClinVar submissions can be increased and because the data are saved in
discrete portable fields, it can be easily extended to future ClinVar submission methods such
as an application programming interface (API).
P35
Copy number variants and regions with absence of heterozygosity in Mexican patients with 45,X Turner syndrome
(1)Laboratorio de Citogenética, Instituto Nacional de Pediatría, CDMX, México. (2)Departamento de Genética Humana, Instituto Nacional de Pediatría, CDMX, México. (3)Inmunogenómica y Enfermedades Metabólicas, Instituto Nacional de Medicina Genómica. CDMX, México. (4)Enfermedades Mendelianas, Instituto Nacional de Medicina Genómica. CDMX, México.(5)Servicio de Endocrinología, Instituto Nacional de Pediatría, CDMX, México.(6)Unidad Genética de la Nutrición, Instituto de Investigaciones Biomédicas, Universidad Nacional Autónoma de México, CDMX, México.
Turner syndrome (TS) is one of the most frequent chromosomal abnormalities in humans,
occurs in 1:2,500 female newborn, TS phenotype includes short stature, gonadal
dysgenesis, heart and kidney malformations, low bone-mineral-density (LBMD), among
others. The presence of these signs varies from patient to patient; even if certain phenotype-
karyotype correlations have been proposed in TS, the clinical characteristics of patients with
the same karyotype can vary. This variability could be related to the presence of copy
number variation (CNV) or regions with an absence of heterozigosity (AOH) in TS patients.
The aim of this study is to find a possible correlation between clinical characteristics and the
presence of CNV or AOH in TS patients with 45,X karyotype.
Previous consent, we collected blood samples from 30 TS patients with 45,X karyotype
without mosaic, searched by interphase FISH and 20 female controls, for all samples gDNA
was obtained and Affymetrix SNP-CN arrays were performed. CNV and AOH were analyzed
using ChAS Affymetrix software. Clinical data were taken from the clinical record. We
focused on clinical manifestations that affect the quality of life as renal malformation,
congenital heart defects, and low BMD to correlate with the CNVs and AOH analysis.
In TS patients and in controls we found some frequent CNV as CN=3 in 14q32.33 and CN=1
in 8p11.22 both considered as benign.
The main findings were: 4 TS patients showed renal malformation, only one of them with
CN=4 in 21q22, region reported with an association with renal malformation, they don't share
AOH regions. 14 TS patients showed congenital heart defects, we found CNVs in 8p23.1,
12p13.31 and 15q11.2 in 7 of them, we found AOH in 11p11.2 in 4 and no one presents
AOH in 22q11.2 reported with an association to congenital heart defects. 17 out of 23 TS
patients present LBMD, none of them presented CNVs in the regions 6p25.1, 20q13.12,
8q22.2 reported with association with LBMD or osteoporosis; 4 of them share CN=3 in
3q22.1, not reported previously. Three TS patients with LBMD presented AOH in 3p21.31
but not in 12q13.11 (VDR locus), reported with an association to LBMD. Our results suggest
that in TS patients the CNV and AOH participate in the variability of clinical features.
P36
Principles guiding prenatal testing in the Belgian genetic centers
Kris Van Den Bogaert1, Nathalie Brison1, BeSHG Workgroup on Prenatal Genetic Testing2,Thomy de Ravel1, Koenraad Devriendt1, Joris Vermeesch1
1 Center of Human Genetics, University Hospitals Leuven, Belgium; 2 The ‘BeSHG Workgroup on Prenatal Genetic Testing’ is composed of members of all Belgian genetic centers
The recent evolution of genomic technologies radically changed the field of prenatal genetic
testing. In 2013, a national consensus between the eight Belgian genetic centers was
reached to use genomic arrays as a first-tier diagnostic test for the detection of chromosomal
aberrations in prenatal invasive samples. Soon thereafter, non-invasive prenatal testing
(NIPT) was increasingly offered for fetal aneuploidy detection demonstrating high
sensitivities and specificities for trisomy 21, 18 and 13. Since July 2017, NIPT has been
reimbursed to all pregnant women in Belgium. As the Belgian genetic centers apply a
genome-wide NIPT approach, other genomic imbalances that are clinically relevant for fetal
or maternal health are detected in ~1% of all samples. These incidental findings include (i)
other fetal aneuploidies, (ii) fetal or maternal segmental imbalances and (iii) maternal cancer.
A national consensus approach is presented on how the interpretation of invasive prenatal
array results as well as the reporting of incidental findings detected by NIPT are managed in
Belgium. In addition, we demonstrate the benefits of sharing prenatal array and NIPT data in
a national database, as this constitutes an elaborate source of data, which can be used for
technical benchmarking or mined for genotype-phenotype correlations. Altogether, we
demonstrate the added value of establishing national consensus guidelines and data sharing
as it shows to improve pregnancy management.
P37
Curation of Metabolic Disease Genes: The ClinGen Inborn Errors of Metabolism Working Group and Phenylalanine Hydroxylase
Diane B. Zastrow1,2, Heather Baudet3, Cindy Si4, Amanda Thomas5, Meredith Weaver6, Wei Shen7, Jixia Liu8, Rachel Mangels2, Jonathan S. Berg3, Stephen F. Dobrowski9, Karen Eilbeck10, Gregory Enns2, Annette Feigenbaum11, Uta Lichter-Konecki12, Elaine Lyon7,10, Marzia Pasquali10, Nenad Blau13, Robert D. Steiner14, William J. Craigen15, and Rong Mao7 for the ClinGen Inborn Errors of Metabolism Working Group
1Palo Alto Medical Foundation, CA, USA; 2Stanford University, CA, USA; 3University of North Carolina, Chapel Hill, NC, USA; 4GeneDx, Gaithersburg, MD, USA; 5Columbia University Irving Medical Center, New York, NY, USA; 6American College of Medical Genetics and Genomics, Bethesda, MD, USA; 7ARUP Laboratories, Salt Lake City, UT, USA; 8Marshfield Clinic Research Foundation, WI, USA; 9University of Pittsburgh Medical Center, PA, USA; 10University of Utah, Salt Lake City, UT, USA; 11Rady Children's Hospital, San Diego, CA, USA; 12Children’s Hospital of Pittsburg, PA, USA; 13University Children's Hospital, Heidelberg, Germany; 14University of Wisconsin School of Medicine and Public Health, Madison, WI, USA; 15Baylor College of Medicine, Houston, TX, USA.
The ClinGen Inborn Errors of Metabolism Working Group was tasked with creating a
comprehensive, standardized knowledge base of genes and variants for metabolic diseases.
