Model fit P-value=0

[贾晓媛]

Hello,

I met a problem when using PRSice2 in my datasets.

In the output, I got some model fit P-values as 0, the following is the results in .prisce file and the bar plot:

Pheno Set Threshold R2 P Coefficient Standard.Error Num_SNP

- Base 5e-08 0.0820902 0 36.0257 0.85187 122

- Base 5e-06 0.0826352 0 51.0744 1.20337 180

- Base 1e-05 0.0826138 0 55.9534 1.31851 199

- Base 0.0005 0.0875756 0 243 5.54652 894

- Base 0.05 0.0508741 4.55297e-229 4005.86 122.354 34747

- Base 0.1 0.0402486 1.13084e-180 5267.54 181.894 61747

- Base 0.5 0.0283772 3.26095e-127 11068.8 458.004 211979

- Base 1 0.0276022 9.5865e-124 15437 647.919 306759

I think it might be due to the extremely small P-values for the model fitting.

Could you please let me know how to solve this problem?

Thank you very much.

Best regards,

Xiaoyuan

[Sam Choi]

That's caused by extremely small P-value (p-value less than 1e-321 or around that). There's nothing that we can do as that's just what it is. You might want to check for overlapped samples as that can cause extreme p-value, though it usually also caused extremely high R2, which isn't observed in this case.