Fig 4.tif (153.54 kB)

CR-specific ARE enrichment of stronger GWAS CRC p-values accounting for known loci and variant correlation.

Download (153.54 kB)
posted on 21.11.2017, 18:41 by Stephanie A. Bien, Paul L. Auer, Tabitha A. Harrison, Conghui Qu, Charles M. Connolly, Peyton G. Greenside, Sai Chen, Sonja I. Berndt, Stéphane Bézieau, Hyun M. Kang, Jeroen Huyghe, Hermann Brenner, Graham Casey, Andrew T. Chan, John L. Hopper, Barbara L. Banbury, Jenny Chang-Claude, Stephen J. Chanock, Robert W. Haile, Michael Hoffmeister, Christian Fuchsberger, Mark A. Jenkins, Suzanne M. Leal, Mathieu Lemire, Polly A. Newcomb, Steven Gallinger, John D. Potter, Robert E. Schoen, Martha L. Slattery, Joshua D. Smith, Loic Le Marchand, Emily White, Brent W. Zanke, Goncalo R. Abeçasis, Christopher S. Carlson, Ulrike Peters, Deborah A. Nickerson, Anshul Kundaje, Li Hsu

Variant-CRC association p-values for CR ARE variants were compared to ARE variants of all non-digestive tissues. The y-axis shows the–log10(KS test p-value) reflecting the significance of the enrichment. Enrichment analyses were performed for single variant-CRC association p-values from GWAS unadjusted for known loci (shown in black) and from GWAS that included a polygenic risk score (PRS) in the model (shown in gray). Results across different LD pruning schemes are shown using three different correlation r2 thresholds. For each threshold, LD blocks were defined as sets of correlated CR ARE variants or ARE variants from non-digestive tissues with r2 greater than or equal to 0.9, 0.8, or 0.5. For each LD block, a priority pruning scheme was employed selecting the ARE variant with the strongest CRC association p-value. Enrichment tests were repeated for each pruning threshold and compared to enrichment results without LD pruning (‘No Threshold’).