Dear Opentargets team,
I am wondering is there any roadmap to integrate TWAS workflow to opentargets platform?
Thanks.
Shicheng
We currently don’t have TWAS on our roadmap. Although TWAS has been successful at identifying genes whose genetically regulated expression is associated to traits, a major limitation of TWAS is that it cannot distinguish between a causal effect of expression on disease and a tagging association within the same region due to correlations among SNPs (i.e. linkage disequilibrium, LD). Hence, it leads to a high number of false positive causal genes results. In contrast to TWAS, colocalization analyses, including COLOC, which is what we use in the Genetics portal, focuses on estimating the probability of the SNPs causals for eQTL to be the same as for GWAS, irrespective of direction of genetic effect on expression or disease, and are not designed to test for an effect of gene expression on disease risk. COLOC output is currently used as a feature for our L2G score. However, if the are clear evidence that TWAS may improve the prediction model for causal genes, we will be happy to investigate further.
1 Like
Thanks Maya. Here are my thoughts on using TWAS with only colocalized eQTL SNPs:
The proposed approach is to first apply coloc to identify causal eQTL SNPs, and then run TWAS using only those coloc-significant eQTLs for the association testing.
The benefits of this approach are that it can strengthen the causal link between gene expression and trait, reduce multiple testing burden, and improve power by focusing on SNPs most likely to be functional. Coloc filtering helps remove non-causal eQTLs.
Hi Jessica,
Fantastic!! I do looking for vendor to provide such service! Please feel free to contact me with sguo@arrowheadpharma.com
Looking forward to some collaboration!
Best,
Shicheng