Is this expected that HaplotypCaller identified 40% more variants by 2.2.5 than 2.1.8

skblazerskblazer Posts: 18Member

Hi GATK team,

I recently downloaded GATK2.2.5 and re-analyzed a small part of my data which was analyzed by2.1.8 pipeline previously, following the same parameters. However I found the new pipeline identified 7247 variants (Q30) while 2.1.8 previously detected 5119. Fortunately, 5118 of the previous ones were recovered by 2.2.5 and 4845 of them were given the same genotypes for all samples. But I still want to know whether the previous version would indeed miss some data.

I checked your release notes, but it just claimed some performance issues from 2.1.* to 2.2.* for HC.



Best Answer


  • Geraldine_VdAuweraGeraldine_VdAuwera Posts: 8,690Administrator, GATK Dev admin

    Hi SK,

    In v2.2 we fixed a number of bugs in HC that are not detailed in the release notes, maybe that's why you're seeing a difference -- but that does seem like a big difference. Have you looked at the "new/extra" calls to see if they have any distinguishing features?

    Geraldine Van der Auwera, PhD

  • ebanksebanks Broad InstitutePosts: 690Member, Administrator, GATK Dev, Broadie, Moderator, DSDE Dev, GP admin

    It's probably related to the bug fixes for the Exact model which are detailed in the release notes.

    Eric Banks, PhD -- Director, Data Sciences and Data Engineering, Broad Institute of Harvard and MIT

  • skblazerskblazer Posts: 18Member
    edited November 2012

    Hi GATK developers,

    After I compare them in details, I found the new pipeline splits most of the complex variants back to two independent SNPs, such as:

    In 2.1.8 pipeline:

    SCF1 343 . TCTATT ACTATC 1117.34 . .

    In 2.2.5 pipeline:

    SCF1 343 . T A 1121.23 . ...

    SCF1 348 . T C 1121.23 . ...

    Is this expected?


    Post edited by skblazer on
Sign In or Register to comment.