SNPs with extremely low P-value

[中杤昌弘]

I would like to pick up lead SNPs and Ind Sig SNPs from my data. But, my data (summary statistics) has some SNPs with extremely low P values like P-value=5.0x10^-700.

Could you tell me whether the FUMA tools can correctly detect lead SNPs and Ind Sig SNPs from my data? 

[Kristen Kelly]

Hello,

As long as the relationships between SNPs in your data still follow the patterns typical for GWAS data (eg. SNPs in closer linkage disequilibrium with something significant also tend to come out with lower p-values) this should probably be OK.

The one potential issue might be related to storage of the value "5.0x10^-700" - a number like that is so small it may get treated as "0" instead of stored using floating-point approximation, and I'm not sure if any parts of FUMA might have a problem handling 0s. Most likely it will still handle it normally, but I suggest double-checking your results against what you see in the data yourself (eg. making sure what's marked as a lead or independently significant SNP matches your expectations) and giving me the job ID after you've run it so I can check the job logs for you and let you know if I see any unusual warnings or errors.

Kristen

PS. If there did turn out to be any problems, you would still be able to process your data through FUMA by identifying lead SNPs using another tool, and uploading a list under "Pre-defined lead SNPs", then unchecking the box "Identify additional independent lead SNPs" so FUMA will skip the step where it identifies lead SNPs. If you use this method, the list of pre-defined lead SNPs should have three columns, one for the SNP rsid, one for chromosome number, and one for position (GRCh37).

[Masahiro Nakatochi]

Thank you for your answer.

I have confirmed whether the lead SNPs with the lowest P-values were picked up in the loci including SNPs with  P=5.0x10^-700. 

As a results, a different SNP from the SNP with the smallest P value was chosen as the lead SNP.

Therefore, I will manually explore the read SNPs.

2020年12月16日水曜日 23:07:40 UTC+9 Kristen Kelly:

[Kristen Kelly]

Hello,

Would you be willing to share the job ID with me? I would like to look into this further and see exactly what FUMA did when processing your data.

There can be normal circumstances where a highly significant SNP doesn't get selected as a lead SNP due to patterns of LD with other highly significant SNPs. But given the potential complication with accurately storing p-values that low, I'd like to have a more detailed look at exactly what FUMA did as it processed your data - if there were any problems, that's something we'd want to find and fix for the future!

Thanks!

Kristen