Hello,
I have a question about reporting low coverage credible sets.
We performed a discovery GWAS of a condition that is quite rare in the population (~1-2%) and selected suggestively significant signals (P<5x10-6) for replication. We then ran Susie on replicated regions. Some of the weaker signals only produced a credible set at a coverage of 55%. In a previous post, you mentioned that this means the credible set has only 55% chance of capturing a non-zero effect. But if we are fairly confident that our signal is real (because it replicated in an independent cohort), is it appropriate to report this credible set and to state that the variants in the credible set are the most likely causal variants? We would of course provide the coverage used in the methods and highlight that other, stronger signals are higher priority for follow up.
I hope this question makes sense.
Thanks very much!
Hello,
I have a question about reporting low coverage credible sets.
We performed a discovery GWAS of a condition that is quite rare in the population (~1-2%) and selected suggestively significant signals (P<5x10-6) for replication. We then ran Susie on replicated regions. Some of the weaker signals only produced a credible set at a coverage of 55%. In a previous post, you mentioned that this means the credible set has only 55% chance of capturing a non-zero effect. But if we are fairly confident that our signal is real (because it replicated in an independent cohort), is it appropriate to report this credible set and to state that the variants in the credible set are the most likely causal variants? We would of course provide the coverage used in the methods and highlight that other, stronger signals are higher priority for follow up.
I hope this question makes sense.
Thanks very much!