Bug Report: mean genotype input (-gmode 1) produce erroneous results

[Trefoil]

Hi BIMBAM developers,

To briefly describe the problem:

I first ran bimbam [latest linux version] on my chip data and use
hapmap as reference panel to simultaneously produce imputed genotypes
and do association test. Then I took the resulting mean genotypes as
input (-gmode 1) but found I could no longer reproduce the previous
association results. During the execution time, screen output said
many SNPs had been removed because their maf<0, and the
resulting .snpdata.txt file indicated that most SNPs have maf
approaching zero; both of which were clearly not the case for my data.
When I tried to feed bimbam with genotype distribution (-gmode 2), no
such problem was found. So I think it is a bug for "-gmode 1" input
mode.

I can send my data to you if necessary to help debugging. Thanks!

Xueya

[Yongtao Guan]

It seems that you used genotype distribution file instead of mean
genotype file for gmode 1.
Please send me the log file if it's not the case.
Thanks,
Grant

--
Yongtao Guan, PhD
Postdoctoral Fellow
Dept of Human Genetics
University of Chicago
http://home.uchicago.edu/~ytguan/