How to identify the important SNPs from several GWAS studies

Question

[senowinski]

When trying to pool all the SNPs identified as being associated with a disease from different GWAS studies. How do you determine which are actually of interest? Are there any other things (than the below) to look out for when determining whether or not SNPs found significantly associated to a disease by GWAS are definitely of interest?

1). Validated in other cohorts

2). Methodologically superior

3). Sample size

4). Found in different populations

5). How significant they were found in the study (ORs & was multiple correction applied)

[GouthamAtla]

Are you asking how reliable the published GWAS SNPs ?

[senowinski]

Yes @geek_y I am asking that exactly. While sometimes the SNP identified by GWAS may be in LD with the "real" SNP, I thought this could be circumvented by identifying those SNPs significantly associated with a disease by several studies? Whereas identifying how other SNPs are reliable, I could use the aforementioned (point 2-5)? Any suggestions? How would you determine whether or not a published SNP was reliable?

Answer 1

[ivivek_ngs]

This review should be able to give you are starting point.

The basic idea in a nutshell is GWAS will give you are lot of SNPs which mostly will be in non-coding regions. Now to understand their functionality and prioritize them you will need to understand their functional impact. You need to make cis-eQTL analysis for this. One such requirement should be furnished by the ENSEMBLE regulatory build. However, I would encourage you to look in these tools as well. All you need to understand is the SNPs found by GWAS (mostly non-coding) have pathogenic relevance or not based on functional prioritization.

Take a look at this FUMAGWAS as well. Also this SuRFR R package can be useful. Hope these can be suitable start points for you.