|
|
Post by brodyj on Mar 31, 2021 21:12:08 GMT
When running --cojo-slct in GCTA some of my input SNPs in the .ma file have very small p-values, < p=1e-700. The results files ( .jma.cojo ) contain zeros for p and pJ for these loci. I wanted to confirm that this is a precision issue with printing the p value, but otherwise the algorithm would be able to run correctly with these small values. Thanks for your help and the very useful package!
|
|
ax
New Member
Posts: 27
|
Post by ax on Apr 26, 2021 8:01:33 GMT
When running --cojo-slct in GCTA some of my input SNPs in the .ma file have very small p-values, < p=1e-700. The results files ( .jma.cojo ) contain zeros for p and pJ for these loci. I wanted to confirm that this is a precision issue with printing the p value, but otherwise the algorithm would be able to run correctly with these small values. Thanks for your help and the very useful package! Hi brodyj, It is the precision issue. It won't affect the analysis as in the algorithm it converts to z-score first. You can use the z-score to convert back to p-value when you have extremely small p-value by specific R function, pnorm(q, mean = 0, sd = 1, lower.tail = TRUE, log.p = FALSE) by turn on the log.p flag to log.p = TRUE Cheers, AX |
|
