Talk 2 Baldock Atlas

8/18/2019 Talk 2 Baldock Atlas

1/83

Atlas-Based Bioinformatics

Richard BaldockMRC Human Genetics Unit Institute of Genetics and Molecular MedicineEdinburgh, UK

Thursday, 16 February 2012


2/83

R A Baldock Dundee 2012 2

eMouseAtlas - Development and Gene-Expression

GENE EXPRESSION

Carol Wicking, University of Queensland



3/83



GENE EXPRESSION

Carol Wicking, University of Queensland



4/83



TRANSGENICS

cre induced-LacZ expression

W. Hsu, University of RochesterAcridine OrangeJ. Burns

University of Bristol

APOPTOSIS

Sigolène MeilhacInstitut Pasteur

CELL LINEAGE

anti-BrdU staining J. McDearmid

McGill University

CELL DIVISION



5/83



TRANSGENICS

cre induced-LacZ expression

W. Hsu, University of RochesterAcridine OrangeJ. Burns

University of Bristol

APOPTOSIS

Sigolène MeilhacInstitut Pasteur

CELL LINEAGE

anti-BrdU staining J. McDearmid

McGill University

CELL DIVISION



6/83


eMouseAtlas Framework



7/83


eMouseAtlas Framework



8/83


VPH

Atlas-Based Infrastructures

emage

Population

Organism

Tissue

Cell

Organelle

Molecule

Basic research Clinical practice

Hudsen Virtual

Patient

Personal

Atlas

Visible

Human



9/83


eMouseAtlas



10/83


eMouseAtlas



11/83


eMouseAtlas



12/83


eMouseAtlas



13/83


EMAGE - current status

~22K submissions

Range Theiler 7-25

! MGI/GXD (~1,300)! Direct submission (~300)

! Large scale screens (~16K)

Repository DB: 38.4K entries

submissionsThursday, 16 February 2012


14/83


EMAGE - current status

~22K submissions

Range Theiler 7-25

! MGI/GXD (~1,300)! Direct submission (~300)

! Large scale screens (~16K)

Repository DB: 38.4K entries

submissionsThursday, 16 February 2012


15/83


EMAGE Embryo Space



16/83


EMAGE Embryo Space



17/83


EMAGE Embryo Space



18/83


BioAtlas - data mapping

• image collection - no mapping all spatialinterpretation left to user

• implicit mapping - data interpreted andannotated with controlled vocabulary or

ontology, image requires interpretation butsome query and pattern analysis possible -

“simple”

• explicit mapping, full spatial delineation ofinformation e.g. expression pattern.



19/83





20/83





21/83





22/83


Spatial Data Mapping

• Manual tie-point alignment (WlzWarp)! mesh-based constrained distance transform

! interactive

! arbitrary complexity

• Automated fine tuning (ITK/ANTS)

• Editor review



23/83


3D Data Mapping - WlzWarp



24/83





25/83





26/83


3D Mapping - Wnt signalling pathway



27/83


3D Mapping - Wnt signalling pathway



28/83


3D Visualisation

• Applications:! SectionBrowser, JAtlasViewer

! Format conversion -> a.n.other

• Browser-Based! canned views & movies

! Tiled zoom-viewer

! Extended to 3D protocol (IIP3D)

! Multi-layer

! Interactive overlays

! WebGL



29/83


OME Project Objectives

To use and extend OMERO to meet mouse atlas andIGMM requirements:

•Embed woolz images

•Sparse reconstruction & mapping•large image data

•Annotation overlay and visualisation

•3D mapping - e.g. OPT images

•IGMM imaging - archiving and analysis



30/83


Woolz images



31/83


Woolz images



32/83


Woolz images



33/83


Woolz images



34/83


Woolz images

• 330K lines Ansi C code

• 184K lines application code



35/83


Woolz Images



36/83


Woolz Images



37/83


Woolz Images



38/83


Woolz Images



39/83


Woolz Images



40/83


Woolz Images

• Arbitrary bounding box

• Interval coding - compact• Fast binary & morphological operations

• Separation of domain from values enables valuedata sharing



41/83


Woolz Images



42/83


Woolz Images

• value types: ubyte, short, int, float,double, rgba, bitmap• Value table can be shared by manyobjects

• Compact coding without compression• Iterators to navigate data• Tiled data can be memory mappedfor very fast access - minimal coding

change



43/83


Woolz Images

• value types: ubyte, short, int, float,double, rgba, bitmap• Value table can be shared by manyobjects

• Compact coding without compression• Iterators to navigate data• Tiled data can be memory mappedfor very fast access - minimal coding

change



44/83


Woolz image objects

• Polylines, boundary lists• histograms

• meshes - 2D & 3D

• transforms! affine

! basis function

! mesh

! conforming mesh



45/83


Woolz image objects




! basis function

! mesh

! conforming mesh



46/83


Woolz image objects




! basis function

! mesh

! conforming mesh



47/83


Woolz image objects




! basis function

! mesh

! conforming mesh



48/83


Woolz image objects




! basis function

! mesh

! conforming mesh



49/83


Sparse Image Reconstruction & Mapping

•EurExpress project

! 19.5K in situ probes, 350K images

! ~24 images per in situ probe

! ~0.5 micron resolution in plane

! 150 micron plane separation

• EmbryoExpress - 20K images

• Allen Brain Atlas - 200K images

• Require automation - Advanced Normalisation Tool (ANTs)! sparse image, matching mask

! Full 3D affine then non-linear warping.



50/83


• Semi-automatic reconstruction

• Automated segmentation

•Manual mapping

• 2D pseudo wholemount

• full 3D in progress

Pseudo Wholemount Mapping to Emap - done


P d Wh l M E d


51/83




•Manual mapping





P d Wh l M i E d


52/83




•Manual mapping





P d Wh l M i E d


53/83




•Manual mapping





3D M i


54/83


3D Mapping to emap


3D M i


55/83


3D Mapping to emap


3D M i t


56/83


3D Mapping to emap


3D M i t


57/83


3D Mapping to emap


3D M i t


58/83


3D Mapping to emap


Large Image Data


59/83


Large Image Data

• Single reconstructions already 30+GB• New EM embryo data ~0.5TB

• OPT data - small (200MB) but many - 5K

• Typical requirement to browse as sections• Require arbitrary angle re-sectioning

• BLB - want to browse online using no more thata web-browser


Tiled Image Servers


60/83


Tiled Image Servers

[maps.google.com]

[Shawn Mikula, Issac Trotts, James M. Stone, and Edward G. Jones,Internet-Enabled High-

Resolution Brain Mapping and Virtual Microscopy, Neuroimage, vol 35(1), p. 11, 2007]


Sectioning Parameters


61/83


Sectioning Parameters

• Angles:

! Pitch, Yaw, Roll

• Position:

! Fixed point (f) &distance

• Scale

• Tiling depends onorientation & scale


IIP3D Extensions


62/83


IIP3D - Extensions


IIP3D


63/83


IIP3D


IIP3DWeb App Architecture


64/83

R A Baldock Optos PLC, 22/3/11 27

IIP3D Web-App Architecture


IIP3DWeb App Architecture


65/83


IIP3D Web-App Architecture

DB image server

Remote Object

protocol


Performance


66/83


Performance


Performance


67/83


Performance


IIP3D Clients


68/83


IIP3D Clients

• Javascript! Ajax

! MVC design, uses MooTools & Yahoo Widgets

• Multi-section at high resolution

• Volume overlays

• Anatomy & gene-expression overlays

• Controls! viewing angles - virtual sections

! zoom

! distance, fixed point

! section locator & view angle feedback

! distance measurement, query by image value

! ......


Annotation Overlay & Visualisation


69/83


Annotation Overlay & Visualisation

• Atlas models include ontology and domains -image regions for anatomical terms. Typicallyexclusive

• Gene-expression data, open ended, multipleoverlapping patterns


Layers Overlays & Indexed Objects


70/83


Layers, Overlays & Indexed Objects

• Multiple layers via html image overlay including opacity! currently layers constrained to identical domains

• Regional overlays using an indexed object! currently using “compound object” - regions can overlap

! TBD - standard indexed volume - regions spatially exclusive


Image Processing IIP3D extensions


71/83


Image Processing IIP3D extensions


IIP3D Examples


72/83


IIP3D Examples


IIP3D Examples


73/83


IIP3D Examples


IIP3D Examples


74/83


IIP3D Examples


IIP3D Examples


75/83


IIP3D Examples


WebGL


76/83


•IIP3D viewer extension to 3D visualisation

! Use X3Dom - Javascript binding to X3D

• navigation feedback

• Anatomy visualisation! see demo

• Extend to 4D


3D mapping - WlzWarp


77/83

R A Baldock Toshiba Medical Visualisation Systems Europe 21/4/11

3D mapping WlzWarp

•Allows placement of landmarks (points of

equivalence) on source and target on volume

renders instead of isosurfaces

• On-the-fly feedback of warping progress

• Uses constrained distance transform (CDT) inwarping• Woolz, Qt, Coin3D(+SIMVoleon)• Linux, OS X, Windoes

• Open Source (Free!)



78/83

3D mapping - WlzWarp


79/83

R A Baldock Toshiba Medical Visualisation Systems Europe 21/4/11

3D mapping WlzWarp

•Allows placement of landmarks (points of

equivalence) on source and target on volume

renders instead of isosurfaces

• On-the-fly feedback of warping progress

• Uses constrained distance transform (CDT) inwarping• Woolz, Qt, Coin3D(+SIMVoleon)• Linux, OS X, Windoes

• Open Source (Free!)


Lineage & the Brain


80/83


g

•Embryo development 7-5-8.5 dpc

• 11-fold growth of ectoderm cell layer

• complex folding

• lineage clones via HRP cell labelling(iontophoresis)

• pattern recognition very difficult

• conformal transform of ectodermal surface to “flat-map”


Lineage & the Brain


81/83


g

A Camus, K Lawson, W Hill et al

Development 2011


Lineage & the Brain


82/83


g

A Camus, K Lawson, W Hill et al

Development 2011


eMouseAtlas


83/83

MRC Human Genetics Unit, Edinburgh

EMAGE:Chris Armit

Shanmugasundaram VenkataramanLorna RichardsonPeter Stevenson

EMAP:Albert Burger

Bill HillNick BurtonYiya YangJulie MossLiz Graham

Allyson Ross

Duncan DavidsonRichard Baldock

Other

Jonathan BardMatt Kaufman

NIH GUDMAP:Simon Harding

Bernard HaggertyKoosum Roochum

BBSRC:Mike Wicks

FP7 RICORDO:Xu Gu

Wellcome OME:Jianguo Rao

University of Edinburgh

Douglas ArmstrongNestor Milyeav

Institute of Human Genetics, Newcastle University

Susan LindsayJanet Kerwin

Colin SemplePedro Coutinho

Ian Overton

Heriot Watt University

Albert Burger

Date post:	06-Jul-2018
Category:	Documents
Upload:	mikhaela-malabonga
View:	223 times
Download:	0 times

Talk 2 Baldock Atlas

Documents