broadinstitute / tensorqtl

Ultrafast GPU-enabled QTL mapper
BSD 3-Clause "New" or "Revised" License
157 stars 51 forks source link

No credible sets output from susie.map because pval_nominal=0 #129

Open cldebban opened 6 months ago

cldebban commented 6 months ago

Hello,

I recently ran Susie fine mapping using tensoQTL's susie.map function. There were several loci where when I looked at the Manhattan plots for the QTL, it seemed like they should have had some sort of credible set, but there was none in the susie.map output. Looking more closely at those loci, it seems that many of the values of pval_nominal are 0 at those loci in the cis mapping output. I'm guessing that this is what was causing issues with Susie.

Do you have any ideas for how to address this?

Thanks, Catherine

francois-a commented 5 months ago

Hi, could you please provide a specific example to reproduce this issue? In principle p-values that are 0 shouldn't pose any problems since p-values aren't used in the calculation.

cldebban commented 5 months ago

Hmm, I'm not sure since I don't think I can share the data -- I'll look into whether I can figure out something reproducible that I can share.

francois-a commented 3 weeks ago

FYI e530163ccbf80abe0d88c46b24519a88aee4f2cc adds the option to return nominal p-values as -log10(P), but that won't affect fine-mapping. Let me know if this is still an issue.