pd3
commented
3 years ago The program usage pages explains how it arrives to its results
Determine sample sex by checking genotype likelihoods (GL,PL) or genotypes (GT)
in the non-PAR region of chrX. The HWE is assumed, so given the alternate allele
frequency fA and the genotype likelihoods pRR,pRA,pAA, the probabilities are
calculated as
P(dip) = pRR*(1-fA)^2 + pAA*fA^2 + 2*pRA*(1-fA)*fA
P(hap) = pRR*(1-fA) + pAA*fA
When genotype likelihoods are not available, the -e option is used to account
for genotyping errors with -t GT. The alternate allele frequency fA is estimated
directly from the data (the default) or can be provided by an INFO tag.The output then shows how is the sex determined: when the score in the 7th column is negative, the program thinks it is a female, otherwise a male.
You can try to check if there is any correlation between the scores and the sex of the known samples. Next step would be to make sure that the correct allele frequencies are used, also the genotype --error-rate option can be tweaked. It is also a good idea to visually inspect the data: if male samples have many hets in regions where there shouldn't be any, then there is not much the program can do.
toyanji
votti
I tried imputing gender of samples with guessploidy. In fact I have the gender status of some samples and others I don't have. I used guessploidy on the vcf file and it predicted all the samples to be females, which was not the case. How to correctly predict the gender from vcf? The output is given below. Would you please help me with this query