Hello Dear community,
I am really in a fix. I did GWAS using array data on discovery cohort and found some significant hits, but my problem is now we don't have a replication cohort because the collaborators who were supposed to provide us with additional participant samples went incognito ;) (the beauty of doing PhD in the covid era). So is there any alternative to replication or is there any better approach to validate my results.
One possibility would be to find, somewhere out there, a trait or a disease that is somewhat "genetically correlated" with your disease or trait, and use that as validation. You would have to search the literature. It wouldn't be a replication per say, as the population and the trait isn't the same, but since they share some genetic etiology (owing to the genetic correlation), it could add some strength to your results provided they validate.
We did precisely that, we had results for a quantitative trait gwas and validated the results from a meta-analysis of case-controls (most of which was published already, and we added some of our own samples). But that's because the q-trait and the (binary) disease were genetically correlated.
If you know of a trait that is somewhat related to yours, and you know there are publications and summary stats available, then there are ways to estimate the genetic correlation with your trait using only summary stats (using LD score regression for example).
The idea is to try to convince that your results are not spurious.
Thank you for the response. Could you please post the link to your study where you have used the approach ?
Thanks