Problem Annotating Manta Tandem Duplications

[apaul7]

I believe I found a bug. When I try to annotate a vcf that contains a tandem duplication called by Manta Duphold fails to add fold changes to the call.

Here's an example:

chr17   41632594        MantaDUP:TANDEM:1:37095:37095:6:0:0     T       <DUP:TANDEM>    39      PASS    END=41633246;SVTYPE=DUP;SVLEN=652;IMPRECISE;CIPOS=-202,202;CIEND=-271,272;AN=6;AC=1     GT:FT:GQ:PL:PR  0/1:PASS:36:86,0,276:19,10

If I change ALT to <DUP> Duphold has no issue annotating

chr17   41632594        MantaDUP:TANDEM:1:37095:37095:6:0:0     T       <DUP>   39      PASS    END=41633246;SVTYPE=DUP;SVLEN=652;IMPRECISE;CIPOS=-202,202;CIEND=-271,272;AN=6;AC=1;GCF=0.471669        GT:FT:GQ:PL:PR:DHFC:DHFFC:DHBFC 0/1:PASS:36:86,0,276:19,10:1.27586:1.02778:1.27586

command using version 0.1.4:
duphold -v test.vcf --drop --threads 1 --output test.duphold.vcf --fasta $REFERENCE --bam crams/sample.1.cram

[brentp]

this should be easy to fix. Would checking that the ALT startswith <DUP or <DEL be sufficient? are there other types to think about?

[apaul7]

That works for me. I'm not sure if it could affect other types of calls.

Only thing I could think about is maybe using the SVTYPE info field instead. Tried skimming through the source to see if that's possible but I don't have experience with nim. But doing that may cause Duphold to run slower because it would have to locate the specific field instead of a column.

4.2 VCF spec has a few examples with <DEL:x so your proposed solution would cover those as well.

[brentp]

would you mind trying this and verifying that it works for your variants of interst? if so, I'll make a new release presently.

(just gunzip, chmod+x and it should be good to go).
duphold.gz

[apaul7]

It works! Thank you for fixing this so quickly!

[brentp]

this is now out in latest release. thanks for reporting and testing.