haplotypecaller indel format

dcittarodcittaro Posts: 31Member
edited September 2012 in Ask the GATK team

Hi there, I've done with a run of HaplotypeCaller on my samples. I'm now analysing everything with snpEff, although I'm doing this "outside" GATK. I had to stop the analysis because a huge number of errors, all dealing with indels, such as:

Error while processing VCF entry (line 580649) :
    chr21   26718345    .   TAATCCTGAGTTTAA TATCCTAAATGTTTAC    943.26  […]
java.lang.RuntimeException: Insertion '-A+AT' does not start with '+'. This should never happen!
    chr21   35260360    .   CATAACAGTTCAT   AGAGACAGAG  425.22  […]
java.lang.RuntimeException: Deletion '+G-TTC' does not start with '-'. This should never happen!

Of course, this is a snpEff error, nevertheless the Indel format looks quite different from what I've ever seen. Consider the first line above: shouldn't it be like

chr21   26718345    .   AT  T   943.26  […]

(I can't resolve the second right now). Any hint is appreciated at this point. I'm writing to snpEff developer for the same reason...

Post edited by Geraldine_VdAuwera on

Best Answer

  • ebanksebanks Posts: 684GATK Developer mod
    Answer ✓

    Hi there,

    The first set of alleles does not reduce to AT/T. It's actually a much more complex substitution (you'll see if you look closely). The Haplotype Caller is doing the right thing here and this error lies solely in the snpEff code, so unfortunately there's nothing we can do to help you here.

    Eric Banks, PhD -- Senior Group Leader, MPG Analysis, Broad Institute of Harvard and MIT

Answers

  • ebanksebanks Posts: 684GATK Developer mod
    Answer ✓

    Hi there,

    The first set of alleles does not reduce to AT/T. It's actually a much more complex substitution (you'll see if you look closely). The Haplotype Caller is doing the right thing here and this error lies solely in the snpEff code, so unfortunately there's nothing we can do to help you here.

    Eric Banks, PhD -- Senior Group Leader, MPG Analysis, Broad Institute of Harvard and MIT

  • dcittarodcittaro Posts: 31Member

    Thanks. I had the answer from the developer, he confirmed that SnpEff cannot handle mixed events. Well, at least you know that snpEff cannot handle (right now) those events, you may want to add some warning in the pages in which you suggest to use snpEff to predict variant effect. d

Sign In or Register to comment.