The Open Targets helpdesk recently received a question from a user navigating the Genetics Portal.
I’m looking at the colocalisation results in the Open Targets Genetics Portal to see whether a gene has genetic and eQTL support for an indication. Sometimes a gene has colocalisation support, but when I click on the gene prioritisation, it’s a different SNP than the GWAS lead variant.
For example, for PCSK9, under Associated studies: colocalisation analysis, Simvastin|treatment/medication code is among the top results. The lead variant in that study is 1_55027911_C_A.
However, the GWAS study colocalisation section on the study-locus page for 1_55027911_C_A, a different SNP — 1_55023869_C_T — is the lead SNP.
How should I interpret this result? Does this mean the gene still has colocalisation support? Should I not expect the same SNP from the GWAS to be the lead variant in the coloc analysis?
Interpreting colocalisation results
The main point to understand about colocalisation is that it is a test for whether two associations are consistent with having the same causal variant — that is, it’s about the pattern of association of all SNPs at a locus, not only the lead variants.
Quite often, two traits (e.g. GWAS and eQTL) can be colocalised when they have different lead variants. This is simply because many variants are in linkage disequilibrium (LD)
— they are correlated — and noise causes one or the other to be more strongly associated in each study.
We don’t show the lead variants for QTLs, these are only used in the colocalisation analyses. For the variants and GWAS mentioned, 1_55027911_C_A was the lead variant in all cases. In this case, 1_55023869_C_T is a tag variant not a lead variant.
The way to interpret this is that the overall association is consistent with one of the associated variants being causal for an effect on both PCSK9 expression and simvastin treatment, though it isn’t necessarily 1_55027911_C_A that is causal.
A note on lead and tag variants
In the Genetics Portal, a lead variant is the variant at a given associated locus with the most significant (smallest) p-value.
A tag variant is the variant that is correlated with the lead variant (r2>0.7) or present in the credible set at a GWAS-associated signal.