I found some strange entries for indels in my VCF file created by the Unified Genotyper. For example:
4 184513470 . TC T 4009 PASS AC=4;AF=0.250;AN=16;BaseQRankSum=1.972;DP=1315;DS;FS=3.466;HaplotypeScore=537.6937;MLEAC=4;MLEAF=0.250;MQ=52.55;MQ0=0;MQRankSum=-10.581;QD=4.55;ReadPosRankSum=-10.128;SB=-3.500e+01;set=variant2 GT:AD:DP:GQ:PL 0/1:230,0:239:99:282,0,5011 0/0:92,0:95:99:0,133,2435
The first sample has genotype 0/1 with a good GQ value. However, according the allele depth field, there is no read supporting the deletion. When I look at the reads using the IGV, I find some reads supporting the deletion for the first sample (and even some for the second one).
Moreover, when I looked at the AD values for SNPs, I noticed the the sum of all AD values is much less than the coverage shown in the IGV. I filtered duplicated reads in the IGV.
Can someone please give an explanation? This link http://www.broadinstitute.org/gatk/gatkdocs/org_broadinstitute_sting_gatk_walkers_annotator_DepthPerAlleleBySample.html explains the difference between AD and DP, but does not help in my case.
Best greetings, Hans-Ulrich