Genetic Analysis in Human Disease
Power of Genetic Analysis
Success stories Age-related Macular Degeneration Crohn’s Disease Allopecia Areata Type1 Diabetes
Not so successful Ovarian Cancer Obesity
Getting StartedQuestion to be answeredWhich gene(s) are responsible for genetic
susceptibility for Disease A?
What is the measurable difference Clinical phenotype
biomarkers, drug response, outcome
Who is affected Demographics
male/female, ethnic/racial background, age
Study Design
Linkage (single gene diseases: cystic fibrosis, Huntington’s disease, Duchene's Muscular Dystrophy)
Families
Association (complex diseases: RA, SLE, breast cancer, autism, allopecia, AMD, Alzheimer’s)
Case - control
Linkage vs. Association Analysis
5M
Linkage Studies- all in the family Family based method to map location of disease
causing loci
Families Multiplex Trios Sib pairs
Staged Genetic Analysis - RALinkage/Association/Candidate Gene
Association Studies – numbers game Genome-Wide Association Studies (GWAS)
Tests the whole genome for a statistical association between a marker and a trait in unrelated cases and controls
Affecteds Controls
Staged Genetic Analysis - RALinkage/Association/Candidate Gene
So you have a hit: p< 5 x10
Validation/ replication
Dense mapping/Sequencing
Functional Analysis
-7
Validation
Independent replication set Same inclusion/exclusion subject criteria Sample size
Genotyping platform Same polymorphism
Analysis Different ethnic group (added bonus)
Staged Genetic Analysis - RALinkage/Association/Candidate Gene
Dense Mapping/Sequencing
Identifies the boundaries of your signal close in on the target gene/ causal variant find other (common or rare) variants
Functional Analysis
Does your gene make sense? pathway function cell type expression animal models
PTPN22: first non-MHC gene associated with RA (TCR signaling)
Perfect vs Imperfect Worlds
Perfect world Linkage and/or GWAS – identify causative gene
polymorphism for your disease Publish
Imperfect world nothing significant identify genes that have no apparent influence in
your disease of interest Now what?
What Happened? Disease has no genetic component.
Viral, bacterial, environmental Genetic effect is small and your sample size
wasn’t big enough to detect it. CDCV vs CDRV
Phenotype /or demographics too heterogeneous Too many outliers
Wrong controls. Population stratification; admixture
Not asking the right question. wrong statistics, wrong model
Meta-Analysis – Bigger is better Meta-analysis - combines genetic data from
multiple studies; allows identification of new loci Rheumatoid Arthritis Lupus Crohn’s disease
Alzheimer’s Schizophrenia Autism
Influence of Admixture Not all Subjects are the same
Missing heritability
Except for a few diseases (AMD, T1D) genetics explains less than 50% of risk. Large number of genes with small effects
Other influences?
Other ContributorsAny change in gene expression can influence disease
state- not always related directly to DNA sequence
Environmental Epigenetic MicroRNA Microbiome Copy Number Variation Gene-Gene Interactions Alternative splice sites/transcription start sites
GWAS- What have we found?
3,800 SNPs identified for 427 diseases and traits
Genome-Wide Association Studies The promise
Better understanding of biological processes leading to disease pathogenesis
Development of new treatments Identify non-genetic influences of disease Better predictive models of risk
and the reality Few causal variants have been identified
Clinical heterogeneity and complexity of disease Genetic results don’t account for all of disease
risk
Pathway Analysis – Crohn’s disease
Personalized Medicine "5P" Health CarePersonalized medicine is:
Predictive: Uses state-of-the-art molecular and diagnostic tools to precisely predict individual health risks and outcomes
Personalized: Is informed by each person’s unique clinical, social, genetic, genomic, and environmental profile
Preventive: Emphasizes wellness and prevention to stop disease before it progresses
Preemptive: Incorporates action-oriented, individualized health planning
Participatory: Empowers each patient to participate in their own care, with coordinated support from their health care team
http://www.dukepersonalizedmedicine.org/what_is_personalized_medicine
Things to remember
You can never have too many samples You can never collect too much information
on a subject
The more you know about the disease and your subjects, the more homogeneous your study will be and the less interference from “population” noise you will have.
Questions True/ False
Association studies are comprised of many multiplex families With 100 randomly chosen polymorphisms and 10,000
diverse human subjects you have a high probability of finding the causative polymorphism for your disease of interest
It’s better to ascertain all of your case subjects in one small town and all of your control subjects in a distant small town so there is no overlap in genetic composition.
The ability to combine data from different large studies to perform a meta-analysis can result in identifying new loci which were not significant in the original studies
If it weren’t for admixture we would not be able to study complex genetics.