Thread by @thatdnaguy

#ASHG21 WL: Wenhan Lu.
quantifying the extent of pleiotropy using rare variant association data in 281k human exomes

#ASHG21 WL: have large scale rare variant association in UKBB exomes.
Preprint - medrxiv.org/content/10.110…
Explorer at - genebass.org

#ASHG21 WL: Interested in pleiotropy. Look at variant-level pleiotropy. 349k vars. 5.4k with > 1 phenotype. 152 associated with > 20 phenotypes.

#ASHG21 WL: Gene-level pleiotropy. 650 genes associated > 1 phenotype. 24 associated with > 10 phenotypes. Example SLC39A8.

#ASHG21 WL: Looked at ICD diseases lumped into 14 domains (by ICD chapter). Diseases within each domain tend to have stronger correlation than between domains.

#ASHG21 WL: Examples LDLR and ADH1B. LDLR associated with hypercholesterolemia. ADH1B associated with drinking behaviors.

#ASHG21 WL: Look at allelic series - variants with different functional impacts may have different levels and strength of pleiotropy.

#ASHG21 WL: Tested 17.9k genes. 35 show allelic series. 8 show pLoF associated with ICD diseases, but missense with traits look blood parameters

#ASHG21 WL: How to quantify pleiotropy? Many associations are across phenotypes, but may be due to correlation. New method to quantify pleiotropic structure in the presence of known phenotypic correlation

#ASHG21 WL: Test for allelic heterogeniety. assumes all variants are independent, no in LD. Null correlated with some constant. Alternative not correlated.

#ASHG21 WL: Currently limited to quantitative phenotypes. Prefilters for MAF < 0.01% [ hmm ]. Remove ultrarare with allele count < 9. 1,117 continuous phenotypes. 211 unique phenotypes. 3.504 unique pairs of phenotypes tested.

#ASHG21 WL: Example of gene ALB related to albumin and total blood calcium. ALB variants associated with both traits. Effect sizes of missense variants not correlated. But effect sizes of pLoF are for the gene (effect on blood albumin and blood calcium).