Entering edit mode
2.1 years ago
austin7923
▴
10
I am trying to perform a post-GWAS functional analysis of disease from fine-mapped SNPs, and then calculate SNPs in LD with these fine-mapped SNPs.
Since statistical fine-mapping methods of GWAS SNPs were able to prioritize a small subset of variants for further testing while accounting for LD structure, does that meaning SNPs in LD with these fine-mapping SNPs were equally causative?
Thank you.
Are you speaking specifically about the SNPs that were selected in the credible set; or the SNPs in LD with the credible set that were nonetheless excluded?
Hi, I am enquiring about the latter, "SNPs in LD with the credible set that were nonetheless excluded". So after they were excluded after the fine-mapping process, are they still useful or (equally causative) to the fine-mapped SNPs? Is it worthy performing further functional annotation on them? Or just the fine-mapped SNPs will do. Thank you.
When a SNP ("rs1234") is not included in the credible set, it indicates that there is sufficient evidence that the association between rs1234 and phenotype is fully explained by other variants. As such, functional annotations are obviated by the lack of association.
However, as with any statistical analysis, there is always the possibility that the dataset was "unlucky" and the credible set does not include all of the causal variants. In this case, i.e., if you are re-analyzing data in light of this possibility (for instance, by introducing a prior on certain types of functional variants), it is valuable to annotate SNPs outside of the credible set.
Hi, thank you for the great explanation. What if the fine-mapped SNPs (not tag SNPs) were obtained with COJO analysis?
Could we say that even if these SNPs could fully explain a phenotype, but still there is a possibility that these credible sets (fine-mapped SNPs) don't include all of the causal SNPs, due to the nature of statistics (unlucky, for example)? And for that it is valuable to annotate SNPs outside the credible set.