![Page 1: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/1.jpg)
11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika
![Page 2: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/2.jpg)
Clustering of GWAS traits
![Page 3: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/3.jpg)
Fine mapping of GWAS using PICS
• GWAS will only associate a locus • Usually the most significant variant is reported • Fine mapping can help to better pinpoint the locus
– For instance using ImmunoChip
• PICS attempts to predict which of the variants is the causal variant – In contrast to simply taking the top association
![Page 4: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/4.jpg)
Example locus
• Top SNP for MS in IFI30 • Previously in MPV17L2
![Page 5: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/5.jpg)
• Also a second very significant hit
• In near perfect LD • No mention of this variant
2 top hits
![Page 6: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/6.jpg)
• GWAS catalog – Database of top GWAS hits for
many diseases – Does not contain ImmunoChip fine
mapping • PICS analysis based on ImmunoChip.
– Top ImmunoChip variant already expect to be often some distance from older GWAS top hits
• No discrimination between added value PICS over ImmunoChip fine mapping
Comparing PICS SNPs to GWAS catalog top hits
![Page 7: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/7.jpg)
• Many PICS SNPs not in LD with GWAS top SNP
• Might simply be because fine mapping picks up independent signal
Comparing PICS SNPs to GWAS catalog top hits
![Page 8: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/8.jpg)
• PICS tries to predict candidate causal variants
• Major assumptions: – Causal variant is a SNP – Causal variant is
present in data • ~10% single candidate
causal SNP
Candidate causal SNP prediction
![Page 9: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/9.jpg)
• Random SNP only frequency matched from same locus – This is very tricky – Major proof in the method
lies in function enrichment • Authors state 90% of causal
SNPs are non-coding
Enrichment
![Page 10: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/10.jpg)
Causal variants tend to occur near binding sites for master regulators of immune differentiation and
stimulus-dependent gene activation, but only 10–20% directly alter recognizable transcription factor binding motifs. Rather, most non-coding risk variants, including those that alter gene expression, affect non-canonical sequence determinants not well-explained by current
gene regulatory models
![Page 11: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/11.jpg)
Causal SNPs map to immune enhancers
Map 6 histone marks by CHiPseq
RNAseq for CD4+ T cell
Combined data with NIH Epigenomics Project and ENCODE for a total of 56 cell types.
![Page 12: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/12.jpg)
Cluster immune cell types based on cis-regulatory element patterns
§ Enhancers are clustered by the cell type-specificity
§ Enhancers of disease variants are correlated with eRNA
![Page 13: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/13.jpg)
Mapping of autoimmune disease PICS SNPs to immune enhancers
Showed enrichment in B-cell and T-cell enhancers
![Page 14: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/14.jpg)
Mapping of autoimmune disease PICS SNPs to immune enhancers
Showed enrichment in B-cell and T-cell enhancers
Note: Enhancers were enriched approx. 2:1 above background Enough?
![Page 15: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/15.jpg)
Criteria of enhancers
immune specific enhancers criteria: (1) number of normalized mean H3K27ac ChIP-seq extended reads/base >4, and
(2) mean H3K27ac in the top fifteenth percentile when comparing immune cells to non-immune cells/tissues.
![Page 16: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/16.jpg)
PICS vs. GWAS
SNPs that do not correspond to a PICS SNP fail to show any enrichment for T-cell enhancers, relative to locus controls
![Page 17: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/17.jpg)
Compare SNPs from 39 human diseases/traits
to 56 cell types
Multiple Sclerosis clusters with other autoimmune diseases based on • Immune cell types • target organ
![Page 18: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/18.jpg)
![Page 19: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/19.jpg)
Disease variants map to discrete elements in super-enhancers
Reminder: super-‐enhancer?
Clustered or contiguous regulatory regions with unusual strong enrichment for the binding of transcriptional co-activators. PoM, S., & Lieb, J. D. (2014). What are super-‐enhancers? Nature GeneXcs, 47(1), 8–12. doi:10.1038/ng.3167
![Page 20: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/20.jpg)
Disease variants map to discrete elements in super-enhancers
PICS SNPs 7.5 fold enriched in CD4 T-cells super-enhancers IL2RA encodes receptor for T-cell stimulation. Distribution in H3k27ac peaks
![Page 21: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/21.jpg)
Disease variants map to discrete elements in super-enhancers
![Page 22: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/22.jpg)
Disease variants map to discrete elements in super-enhancers
Conclusion: “Super-enhancer may comprise multiple discrete units with distinct regulatory signal, function and phenotypic association ”
![Page 23: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/23.jpg)
Causal variants map to regions of transcription factor binding
Hypothesis: SNPs alter transcription factor binding. PICS SNPs coincide with nucleosome-depleted regions. 31 transcription factors with binding sites higher enriched for PICS SNPs.
![Page 24: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/24.jpg)
Causal variants map to regions of transcription factor binding
Does causal variant disrupt or create sequence motif recognized by TF? Selecting consensus motif from databases. Match with 853 highest probability non-coding PICS SNPs. Comparing with random SNPS in same region. 7% of PICS SNPs affected one of the motif.
Estimation that 11% of non-coding GWAS hits can be attributed to direct disruption or creation of TF binding motif.
![Page 25: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/25.jpg)
Causal variants map to regions of transcription factor binding
Rs17293632 increase risk of CD by AP-1 modification in SMAD3. Alternative TF motif database, don’t show significant results.
![Page 26: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/26.jpg)
Conclusion: “Casual non-coding SNPs modulate transcription factor dependent enhancer activity (and confer disease risk) by altering adjacent DNA bases whose mechanistic roles are not readily explained by existing gene regulatory models ”
Causal variants map to regions of transcription factor binding
![Page 27: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/27.jpg)
eQTLs
• “Incorporation of eQTL SNPs allowed us to link causal non-coding disease variants to specific genes” – this is not new
• Only ~12% of non-coding “causal” SNPs also score as eQTL
![Page 28: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/28.jpg)
GWAS vs eQTL
![Page 29: 11/06/2015 Arnau Vich Vila Patrick Deelen Paranita Ferronika](https://reader033.vdocuments.net/reader033/viewer/2022051405/58a039991a28ab5c588b8109/html5/thumbnails/29.jpg)
• H3K27ac mark in many cell types
• Also DNase1 hypersensitive
• However only degenerate motive match for MEF2C
Position of “causal” eQTL + MS GWAS SNP