|
Post by carolynjumper on Mar 23, 2014 10:09:23 GMT
When running the command
gcta64 --bfile input --grm-cutoff 0.025 --make-grm --out out,
GCTA accepts the --grm-cutoff argument and repeats it in the log file, but does not apply a GRM cutoff. In contrast, when given the very similar command,
gcta64 --grm input --grm-cutoff 0.025 --make-grm --out out,
GCTA does apply the GRM cutoff. I did not see any documentation that the --grm-cutoff flag is not valid when the input is a bfile. Regardless, the behavior is extremely surprising since the GCTA log repeated the --grm-cutoff flag as if it was applying the cutoff. My misunderstanding of this issue cost me many hours of lost work, since I did not realize I was computing heritability without a GRM cutoff until I loaded one of the GRM matrices into python.
I would request that either GCTA would have the option to apply GRM cutoffs during GRM creation, or that giving the --grm-cutoff flag during GRM creation would cause the program to fail with an invalid flag error message. Although I now know about this issue, I would expect that other users are likely to have the same misunderstanding as I did.
Thank you, Carolyn Jumper
|
|
|
Post by Jian Yang on Mar 26, 2014 11:23:17 GMT
Thanks. This will be fixed in the next version.
|
|