variant analysis using Freebayes
0
0
Entering edit mode
9 weeks ago
bitpir ▴ 210

Hi all,

I ran freebayes using the following command line:

##freebayes -b <bam> -f hg38.noalt.fa -v <vcf> --min-alternate-count 3 --min-alternate-fraction 0.2 -t b.bed"


One of the samples have this following output:

chr1    179109038   .   A   C   0.014237    .   AB=0.235294;ABP=13.3567;AC=1;AF=0.5;AN=2;AO=4;CIGAR=1X;DP=17;DPB=17;DPRA=0;EPP=11.6962;EPPR=3.17734;GTI=0;LEN=1;MEANALT=1;MQM=60;MQMR=60;NS=1;NUMALT=1;ODDS=5.71882;PAIRED=1;PAIREDR=0.846154;PAO=0;PQA=0;PQR=0;PRO=0;QA=108;QR=460;RO=13;RPL=4;RPP=11.6962;RPPR=3.17734;RPR=0;RUN=1;SAF=4;SAP=11.6962;SAR=0;SRF=0;SRP=31.2394;SRR=13;TYPE=snp;technology.ILLUMINA=1    GT:DP:AD:RO:QR:AO:QA:GL 0/1:17:13,4:13:460:4:108:-4.86964,0,-36.5987


And the other one has this:

chr1    179109038   .   A   C   51.8359 .   AB=0.448276;ABP=3.68421;AC=1;AF=0.5;AN=2;AO=13;CIGAR=1X;DP=29;DPB=29;DPRA=0;EPP=31.2394;EPPR=3.55317;GTI=0;LEN=1;MEANALT=1;MQM=51.4615;MQMR=60;NS=1;NUMALT=1;ODDS=11.9357;PAIRED=1;PAIREDR=0.9375;PAO=0;PQA=0;PQR=0;PRO=0;QA=343;QR=601;RO=16;RPL=13;RPP=31.2394;RPPR=3.55317;RPR=0;RUN=1;SAF=13;SAP=31.2394;SAR=0;SRF=2;SRP=22.5536;SRR=14;TYPE=snp;technology.ILLUMINA=1  GT:DP:AD:RO:QR:AO:QA:GL 0/1:29:16,13:16:601:13:343:-19.822,0,-45.6764


This variant is a false-positive when compared to NA12878 benchmark.vcf. I am unclear as to how one sample has low QUAL and the other a high QUAL. The BQ and MAPQ for these 2 samples are very similar -- 35 and 57, respectively. Per Freebayes best-practices, I should do vcffilter based on the QUAL, but I would like to understand more what contributes to the QUAL score in this situation?

Any idea would be greatly appreciated! Thanks!

calling freebayes false variant positives • 220 views
0
Entering edit mode

We know that false positive issue happens in variant caling tools. You can read benchmarking papers about this. Nevertheless, I would consider AD and DP field in vcf file as well. Maybe you should filter your vcf using these definitions (DP, MAPQ) instead QUAL.