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!

Mutect1 -> depth of 'PASS' variants


I have used Mutect1 (tumor-control) for calling variants. As mentioned here:
http://archive.broadinstitute.org/cancer/cga/mutect ; that "We currently use cutoffs of at least 14 reads in the tumor and at least 8 in the normal (these cutoffs are applied after removing noisy reads in the preprocessing step)". This means that the variants I get as a result of the analysis should have a depth of at least > 14 in tumor and > 8 in control. However, I do see variants in my resulting vcf file that have DP less than the above mentioned thresholds. I would like to know the explanation for these. Does mutect still calls them 'PASS' based on LOD; or something else should be taken care off here?

Best Answers


  • Dear @Sheila

    -Thanks for your response.
    I do have an example to show. Attached are records ('PASS' variants) for Mutect1 and Mutect2 for the same tumor-normal sample pair of mouse using mm10. We can see for Mutect1 and Mutect2 the DP is less than 14 in tumor in these records. I am not sure if Mutect2 also does consider the same DP thresholds for tumor and normal samples. I have also attached the scripts that I used for both programs.
    Also, for positions: 67152443, chr 1 and 43593584, chr 10; both programs give a different DP (but not less than 14). I calculate DP for Mutect2 as counts of REF + counts of ALT.

    Could you share your views on why this is happening. Also, If I use Mutect2, am I supposed to then apply additional read count filters to the final output?, as right now I used only the default calling.

    Thanks for your time!
    yashna paul

  • Thank you Sheila. This was helpful! @Sheila

Sign In or Register to comment.