Entering edit mode
3.1 years ago
KL_STKLBK
•
0
Hi, I would like to genotype multiple individuals at once to generate one single VCF file with BCFtools 1.10.2. For further downstream filtering it would be important to retain the DP for each individual. After reading the manual, I think the -g (--gvcf) switch should do the trick, i.e.
bcftools mpileup -Ou -f myref.fasta ind_a.bam ind_b.bam -a DP -g 0 | bcftools call -Ov -g 0 -m -o test_call.vcf
So, is this correct?
Also, when I look at the output, I get something like this:
scaffold_1 141 . C T 242 . DP=9;VDB=0.92727;SGB=-1.11282;MQSB=1;MQ0F=0;AC=4;AN=4;DP4=0,0,7,1;MQ=60 GT:PL:DP 1/1:131,12,0:4 1/1:138,12,0:4
Suggesting that the total DP is 9, but when I combine the DP fields of the two individuals, I end up with a total DP of 8. Why would that be?
DP=9 is the INFO/DP which is the total raw depth of the locus. Sample specific DPs are filtered depth i.e. they only consider reads that pass all the in-built read filtering a variant caller uses to detect variants. It is also referred to as FORMAT/DP. The tool must have dropped a read which can be due to any reason such as lower mapping quality, duplicate marked, etc.
The metadata in the header section of your VCF (or gVCF) will have definitions of each metric present in the file.