Investigation of Synthetic Association in GWAS using PheWAS and Exome Sequencing. L. Bastarache, J. Bochenek, T. Edwards, Y. Xu, J. Pulley, E. Bowton, H. Mo, W. Wei, L. Wiley, D. Roden, J. Denny Vanderbilt University, Nashville, TN.

   Genome-wide association studies (GWAS) have identified thousands of common variants associated with human diseases or traits. The biological effects on traits of common GWAS SNP associations are often not apparent by functional annotations. The synthetic association (SA) hypothesis proposes that one or more rare causal alleles on a haplotype containing a common, neutral allele, may explain some GWAS findings. To test the plausibility of this hypothesis, we investigated rare variants located near known GWAS findings. In this study, we evaluated 29,722 European-ancestry individuals genotyped on the Illumina HumanExome array with traits defined by electronic medical records in a phenome-wide association scan (PheWAS). For results reported in the GWAS catalog that reach genome-wide significance (p<5x10-8), we mapped the disease or trait to a PheWAS phenotype. We then evaluated all available missense or nonsense SNPs with minor allele frequency (MAF) 5% in the GWAS-reported genes for that phenotype. We tested 104 unique phenotypes and 1,743 unique phenotype-gene pairs using single SNP tests of association adjusted for age, sex, and principal components. There were 84 associations for rare variants with p<0.01 and 59 that had an odds ratio (OR) greater than 2 or less than 0.5 for 38 unique disease phenotypes. There were 22 associations that exceeded a Bonferroni correction for tested phenotype-gene pairs. For example, with primary biliary cirrhosis, we observed associations with rs200568391 CLEC16A (MAF 0.1%, OR=11.19, p=3.6x10-8) and with IKZF3 rs149299224 (MAF 0.3%, OR=6.24, p=8.7x10-7); the ORs in the GWAS catalog were 1.29 and 1.44, respectively. Similarly, rare variants in genes associated with schizophrenia (ARL6IP4), Crohns disease (NOD2, ATG16L2, C7orf72), hypothyroidism (C6orf15), and ulcerative colitis (C6orf223) had rare variants with p<10-4 and OR=1.82-12.53. Associations with Sjögren's syndrome (COL11A2), esophageal cancer (CORO2B), and basal cell carcinoma (PADI4) were statistically significant. Conditional analysis is ongoing to investigate SA in cases where we genotype the GWAS identified SNP. This study demonstrates that rare variation may contribute to the genetic risk of a wide variety of phenotypes, and that large EMR cohorts and the exome array may accelerate investigation of SA and related mechanisms by which rare variants contribute to risk in heritable traits.

You may contact the first author (during and after the meeting) at