We've moved!
This site is now read-only. You can find our new documentation site and support forum for posting questions here.
Be sure to read our welcome blog!

DP and AD

I wonder if it is correct for Haplotype Caller that DP is filtered coverage and AD is unfiltered coverage. I read in the GATK forum that AD could even be larger than DP.

Best Answer


  • dayzcooldayzcool Member

    Hi Sheila,

    It is still confusing to me.
    It is said that "The annotation used by HC for FORMAT-level DP is DepthPerSampleHC, not Coverage." according to /3266/depthpersamplehc thread.

    Also, I read that "sum(AD) = DP" that according to a comment in the DepthPerSampleHC code.

    For this reason, I think FORMAT-level DP is filtered. If so, INFO-level DP is filtered because DP values were consistent in both FORMAT-level and INFO-level in my observation.

    However, it is confusing to me to read that sum(AD) = DP, because it does not make sense to me that sum(unfiltered AD) is equal to filtered DP. Additionally, I found that sum(AD) <= DP in some vcfs from HC 3.2 I have. I couldn't find an example to confirm that AD could be bigger than DP.

    Thank you!

  • dayzcooldayzcool Member

    Thanks a lot for your kind explanation and writing documentation for it!

Sign In or Register to comment.