Freebayes occasionally incorrectly genotyping large deletions

[Coryza]

Hi,

I'd like to report a small bug that I occasionally see happening in my data-set;

• Extremely large deletions are often called as homozygous while clear heterozygosity can be seen in read-mapping. In the example below you'll see one heterozygous deletion (left, 540 bp) and one homozygous deletion (middle, 314 bp). Based on the supplementary read alignments one can see that these are clearly both heterozygous. So what's going on with the homozygous one?.
  

# The VCF entry of the 540 bp (correct);
Chr03  12661669        .       TAACCAAAAAAAAAATTTTCCAAGTAAAATATTGACAGGTCAAACTTATATGGAATGAAAACCAGAACAACTTGTTTGAACTCTGTAATGGTTATTAGTTACAGAATTTGTTAAATAGACAACCACATCCAACACACTCTGGTCACCAATGATTAGACGTCTGTTGTTAGATAGGGCTATGTTTTCCCTCAAATCCATAAATAGACATACAAACTGCCCAATTTTGTTAGTATCCATAATACATCAAGTCCAGAAAAGTCATACTGCAACACAACACTAACACCACCTATGTTCGATGTTTTCCATAGAAATCTCCACAAAAGTAACATTAAGGATTATTTCATTTTAGCCTGAAAAATTTATAGCAAAACAAAAGGAAGTAATGTCTCTTAAAGTATAGTTAGCCTGAGGCAAAGCAGGAAGCAAATTATGAAAGCAAGAAACTGCATCAACCGAAGCACCCAGAATCACAAAGTGTGGCCACACAAAAATGCACAGGGAATTAATTAGTTGAACCAGCCATAAAAAACGATGAGAAAACAAA  TAAT    481.285 .       AB=0.293103;ABP=24.5753;AC=2;AF=0.666667;AN=3;AO=17;CIGAR=1M540D2M1X;DP=58;DPB=39.5092;DPRA=0;EPP=6.20364;EPPR=3.0103;GTI=0;LEN=4;MEANALT=36;MQM=60;MQMR=60;NS=1;NUMALT=1;ODDS=3.2152;PAIRED=0;PAIREDR=0;PAO=0;PQA=0;PQR=90;PRO=2;QA=637;QR=81;RO=2;RPL=10;RPP=4.1599;RPPR=3.0103;RPR=7;RUN=1;SAF=10;SAP=4.1599;SAR=7;SRF=2;SRP=7.35324;SRR=0;TYPE=complex        GT:DP:AD:RO:QR:AO:QA:PS 0|1|1:58:2,17:2:81:17:637:12542617

# The VCF entry of the 314 bp (incorrect);
21-PSC-MAC_Cavendish.1.0_Chr03  12662590        .       AGCATAATAGTAGAGGGACTAGAGTTCATCCTGAGTACATTTGGACTACGGAGTACGGACTACCTCTCTTCTTCAGTATACTGTGCCCCAGCACTTCTAGCCTTTTCCTCCAGAGTATCATTGGACTCTTTTCACCCCAGTTTGGTGGGGGGGGGGGGGGGGGGGGGGGGCGGGGTGGTCTGAGTTCTCAGCATCACTAAAACCATCACGTTCAGTGCAATTCCATTAAGCATATTAATTTGTGGTTGTGTAAAGTTAATTCCACAAACAGAAATCCCCTTCTGATCTTTTGTCCTATATCTGCTTAACTACAATG AG      91.0037 .       AB=0;ABP=0;AC=3;AF=1;AN=3;AO=18;CIGAR=1M314D1M;DP=62;DPB=42.7563;DPRA=0;EPP=10.7311;EPPR=0;GTI=0;LEN=314;MEANALT=42;MQM=60;MQMR=0;NS=1;NUMALT=1;ODDS=18.7964;PAIRED=0;PAIREDR=0;PAO=0;PQA=0;PQR=0;PRO=0;QA=147;QR=0;RO=0;RPL=12;RPP=7.35324;RPPR=0;RPR=6;RUN=1;SAF=13;SAP=10.7311;SAR=5;SRF=0;SRP=0;SRR=0;TYPE=del   GT:DP:AD:RO:QR:AO:QA 1/1/1:62:0,18:0:0:18:147

Any suggestions? Unfortunately I'm unable to share any of the data due to confidentiality. I'm using Freebayes version v1.3.6 with settings -C 7 -p 3.

For the rest Freebayes works amazingly well on my noisy long read data :)

[Coryza]

Likely similar to #723