Mutect2 paired tumor/normal must be same design ?

Hi GATK team !

While trying to test MuTecT2 intensively on some of my samples, I was wondering about the requirements that you would make on the pair tumor/normal.

For example I have here some normal samples that have are capture-sequenced on a panel of 30 genes.
However my tumor samples come from amplicon sequencing on only 2 genes (included in the 30 above, obviously).
For this amplicon design, I trim the primers and do not mark duplicates.

However, could that (the difference in design and the difference in target size) mislead the results in any way ? Are there some parameters that I should adjust ? (minPruning ? kmers ? others ?).

Thank you for your answer.

Issue · Github
by Sheila

Issue Number
Last Updated
Closed By


  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Having data from very different technologies for the normal vs the tumor is really not ideal, since they will be subject to different technical artifacts. You can get away with it if you have a panel of normals that were generated with the same tech as your tumor sample, but if you don't have that you risk having a lot of false positives due to artifacts. There is no preset method available on the program to deal with that, I'm afraid.

  • manon_sourdeixmanon_sourdeix FranceMember

    Thank you for the answer !
    The amplicon tumor design is very specific for the tumor properties and I do not have a similar calling for normal samples. This is something that users will know and acknowledge, thank you.


Sign In or Register to comment.