"-nan" GP and genotype coding in males

[rcheng]

It is not a bug but I wonder if there is an option for users to choose coding in a desired way. For instance, a male has GT:PL:DP:AD:GP:GQ = 1:180,0:8:0,8:-nan:127 in chrX.

• GP is "-nan" while it seems to be a definite value like "0,1".
• It will allow us to run downstream analyses (e.g. Beagle imputation) more conveniently if GT is coded as "0/0" or "1/1".

[rcheng]

The following GT:PL:DP:AD:GP:GQ are observed in a male on chrX 1:230,235:38:22,16:-nan:127 0:20,38:5:3,2:0.940634,0.0593659:12 0:35,64:13:9,4:0.99192,0.00808017:20 Notes: 1) PL is not normalized; 2) Both REF and ALT alleles exist, e.g. AD = 22,16; 3) GP is observed to be -nan. Wanted to make sure variant calling is alright for males and chrX.

[pd3]

What program was used to generate the VCF?

[rcheng]

The vcf file was generated by bcftools call

[rcheng]

By the way, bcftools call -S worked well using a samples-file. Unfortunately, bcftools convert -S and bctfools view -S failed if the same samples-file was supplied. For instance, bcftools view -S spit an error message "Sample name mismatch: sample #1 not found in the header".

The first line of the samples-file is as follows 4512-JFI-0335 M

Part of the line that contains sample names in the vcf file is as follows #CHROM POS ID REF ALT QUAL FILTER INFO FORMAT 4512-JFI-0335 4512-JFI-0336 4512-JFI-0502 4512-JFI-0503 4512-JFI-0504 4512-JFI-0505 4512-JFI-0506 4512-JFI-0507 4512-JFI-0508 4512-JFI-0579 4512-JFI -0580 4512-JFI-0581 4512-JFI-0582 95-35734 95-38313 95-38314 95-39664 95-39665

We see the first sample in the vcf file is the the sample in the first line of the samples-file.

[pd3]

Can you please provide a small test case to reproduce both problems? Ideally, can you open a new issue for the other one, as it is unrelated to this one?

[rcheng]

Code (testBcftools.s) and data are available at https://palmerlab.s3.sdsc.edu/minio/debug/. Both "-nan" and the "bcftools call -S" error can be reproduced.