I am getting some of my SNPs significant for a phenotypic trait but they are not in LD with their nearby SNPs. My question is that these SNPs which are significant but not in LD with their left and right SNPs are important for my further association work or not? What should I do?
•
written
by
kumar.vinod81
See papers by Lai, Ordovas, et al on the APOA5 gene in humans. A key SNP in this gene is not in LD with its nearby SNPs and shows some very pertinent associations to triglyceride levels. Furthermore, the associations are modified by dietary factors, such that the risk allele does not show as risk until an aspect of the diet passes a certain threshold. In this case, I must respectfully disagree with swbarnes2.
It is also possible that your SNP is tagging a CNV (copy number variant), which other SNPs do not, and it is that CNV that is the causal variant. That was originally thought to be the situation for the lead GWAS signal at IRGM and Crohn disease risk, but that SNP turned out to alter a miR binding site within the 5'-UTR. Nonetheless, that your SNP tags something else is a possibility to consider.
Lastly, what is in LD in one population may not be so in another.
Its a great reply of my question with example. I'll read the above said paper and try to increase my understanding related with LD. Thanks
you mean that LD is needed if a SNP is significant? If it is not SNP is not genotyped in accurate way.....
"needed" is probably too strong of a word, at least in my opinion. With SNP studies, you're typically going to be left with a bunch of SNPs that are statistically associated with whatever you're studying. The question, then, is how to prioritize your validations (be they directed sequencing with another technology or functionally trying to determine the effect of the SNPs). There are a number of ways to prioritize SNPs, one of which is to use LD. The reality is that just because a SNP isn't in LD, that doesn't absolutely mean that it's not worth following up, but perhaps followup should just not be as highly prioritized as those SNPs in LD with their neighbors (if there are any). Basically, the idea is to use whatever data you have to minimize wasting your time following up on SNPs that will turn out to be meaningless.
Thanks for a descriptive reply. I got it and will try to work with on it for my future SNP studies related with SNP prioritization.
Log in to answer this question.
