6/5/2014
Uncovering MS Susceptibility Pathways from Large‐Scale Genetic Studies Chris Cotsapas Yale Neurology/Genetics Broad Institute [email protected]
Multiple sclerosis genetics ca. 2002
• MS risk is heritable (sib risk ~8-14) • HLA DRB1*1501 known for 2 decades – cf. APOEe4 in Alzheimer’s disease • No other genes known • Very large pedigree-based studies were uninformative
1 6/5/2014
And then….
15,000 47 new hits Immunochip
10,000 25 new hits WTCCC2 8000
MS 6000 Subjects 3new hits Meta‐Analysis 4000 Meta v2.5 6new hits 1 new hit 2000 IMSGC GWAS 2 hits ANZ GWAS IMSGC NEJM 2007
De Jager et al. Nat Genet 2009
2007 20082009 2010 2011 2012 Rubio et al. Nat Genet 2009 Date of completion IMSGC Nature 2011 Patsopoulos et al. Ann Neurol 2011
The latest view –97 hits
IMSGC, Nat Genet 2013
2 6/5/2014
… and that’s not all! Exome chip Meta‐analysis v3 + Replication 38K MS cases / 55K controls 34K MS cases / 44K controls 100+ new loci
15,00 47 new hits 0 Immunochip
10,00 25 new hits 0 WTCCC2 800 0 MS Subjects 600 0 3new hits Meta‐Analysis 400 Meta v2.5 0 6new hits 1 new hit 200 IMSGC GWAS 0 2 hits ANZ GWAS
2007 20082009 2010 2011 2012 Date of completion
What on earth do we do now?
3 6/5/2014
The next five (??) years
• More genetic studies – More GWAS – Sequence extremes – Low frequency variation in common disease – Rare diseases • Patient heterogeneity – Progression/severity studies – Therapy • Risk/outcome prediction • Underlying biology
Supplementary File
Supplementary Figure 51. Discovery phase rs6677309. ALet’s zoom into a locus (CD58)
B IMSGC, Nat Genet, 2013
4 6/5/2014
… andSupplementary another File Supplementary Figure 42. Discovery phase rs12946510. A
B IMSGC, Nat Genet, 2013
What we know now
• ~800 risk loci projected – None is necessary or sufficient for disease • Several immune tissues likely involved
5 6/5/2014
What we know now
• ~800 risk loci projected – None is necessary or sufficient for disease • Several immune tissues likely involved
What we know now
• ~800 risk loci projected – None is necessary or sufficient for disease • Several immune tissues likely involved • Do these act independently?
6 6/5/2014
New pathway tools
InWeb, ~12K proteins, ~127K interactions. Lage et al Nat Bio 2007; Rossin et al PLoS G 2011
TNF TNXB Strict significance: LY6G6F AGER EHMT2 p < 5 x 10‐6
EPS15L1
GRB2 BAT2 HSPA1B BAT3
KPNB1 SOCS1 HSPA1A
DOM3Z HSPA8 IL2RA
HSPA1L
VARS Clustering Genetic LSM2 Node # Edge # POLR2C coefficient score
CCHCR1 RDBP
ZBTB9 GTF2H4
P < 0.05 for all network metrics
7 6/5/2014
TNF TNXB
DHX16 EHMT2 EPS15L1 BAT2 SLC39A7 LY6G6F FDR significance: ‐4 GRB2 p < 1 x 10 AGER PTRF CD40 VARS DOM3Z
GTF2H4 SOCS1 IL22RA2 RDBP STK19 HSPA8 CXCR4 CDK4 POLR2G BAT3 IL12B IL2RA POLR2C HSPA1B ZBTB9 HSPA1A KPNB1 CCHCR1 STAT3 HSPA1L LSM2 HSP90AB1 DAXX SKIV2L DDX6 PHGDH NPEPPS ZNRD1 CSNK2B ALDOA Clustering Genetic coefficient score TBKBP1 AGPAT1 Node # Edge # BRD2
DDR1
P < 0.05 for all network metrics
Finding relevant tissues
Network gene expression in a Network gene expression in an random ENCODE tissue immune ENCODE tissue
8 6/5/2014
A pathway burden model
Disease risk
Identifying patient subsets
Pathway 1Pathway 2
Different outcomes?
9 6/5/2014
Acknowledgements
• IMSGC • Networks – Phil De Jager – Kasper Lage – Nikos Patsopoulos • IIBDGC – Jake McCauley – Mark Daly – David Hafler – Hailieng Huang – Adrian Ivinson • Lab folks – Many, many others – Jinmyung Choi • Epigenomics – Parisa Shooshtari – John Stamatoyannopoulos – Mitja Mitrovic – Lizzy Rossin – Ioanna Pagani – Boel Brynedal – Jason Vander Heiden
10