Missing key SEQ for <INS>

[brentp]

Hi, I am trying to use paragraph to genotype manta calls. I get this error:

  Exception: Missing key SEQ for <INS> at 1:145235303; 

and that variant is indeed:

1       145235303       .       T       <INS>   0       .       END=145235303;SVTYPE=INS;CIPOS=0,14;CIEND=0,14;HOMLEN=14;HOMSEQ=GGAT
ATCAGGTTTT;LEFT_SVINSSEQ=GGATATCAGGTTTTCCTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTAAAGTGTGTGGGTCTTT;RIGHT_SVINSSEQ=GACCTCGTGATCCGCCTGCCTCGG
CCTCCCAAAGTGCTGGGATTACAGGCGTGAGCCACCGCGCCCGGCC

so it has left and right, but not SEQ. Is there something I can do to have paragraph genotype these? For example setting SEQ = LEFT_SEQ + N*100 + RIGHT_SEQ ?

[brentp]

Bump. I'm wondering if setting SEQ = LEFT_SVINSSEQ + N*100 + RIGHT_SVINSSEQ
will work with the machinery in paragraph. I can preprocess the VCF to do this but want some idea that it does not violate some internal assumptions.

Is there also some way to genotype BND elements?

thanks!

[atongsa]

what is a homeseq in your info

[atongsa]

vim paragraph/lib/python3/grm/vcfgraph/vcfgraph.py

edit this block to your insert SEQ is ok

        if alt == "<INS>":

[traxexx]

When the full insertion sequence cannot be assembled, Manta will report the sequence near breakpoints instead. Paragraph only uses sequences around breakpoints for genotyping, so @brentp solution should work.

[aksenia]

Nice hack!
vcf.info["LEFT_SVINSSEQ"] + ('N' * 200) + vcf.info["RIGHT_SVINSSEQ"]

will throw an error though, as vcf.info["LEFT_SVINSSEQ"] is returned as a tuple ("ACC...T",), so vcf.info["LEFT_SVINSSEQ"][0] is what will be needed to get to the seq.

[brentp]

Indeed. I am accumulating all the changes here: master...brentp:bp-dev