How the number of Actual calls made can be greater than Confidently called bases ?

sarkarsarkar SwitzerlandMember

Hello,

I am working with de novo assembly and I aligned reads to the scaffolds and called SNPs across the samples. But the result I received after Calling for all confident sites (using Unified Genotyper) is quite unusual to me.

This is what I get:
INFO 22:57:01,315 UnifiedGenotyper - Visited bases 1073801052 INFO 22:57:01,339 UnifiedGenotyper - Callable bases 977317465 INFO 22:57:01,349 UnifiedGenotyper - Confidently called bases 575018059 INFO 22:57:01,357 UnifiedGenotyper - % callable bases of all loci 91.015 INFO 22:57:01,366 UnifiedGenotyper - % confidently called bases of all loci 53.550 INFO 22:57:01,374 UnifiedGenotyper - % confidently called bases of callable loci 58.836 INFO 22:57:01,382 UnifiedGenotyper - Actual calls made 698376287
My question is how the number of Actual calls made is greater than the number of Confidently called bases as I always expect Confidently called bases would be greater or equal to Actual calls made but not less.

I would appreciate your help.
Many Thanks.
sarkar

Answers

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Hi Sarkar,

    I'm not sure -- can you tell me which version of GATK you are using?

  • sarkarsarkar SwitzerlandMember
    edited November 2013

    Hi Geraldine,
    I am using GenomeAnalysisTK-1.2-4-gd9ea764. This version in installed in the servers of Swiss Institute of Bioinformatics.

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    Oh, that's a really old version, and is probably full of bugs. I strongly suggest you upgrade to a more recent if not the latest version. If you don't have the permissions to update software yourself, please contact your systems administrator to get them to do it for you. You will get much better results.

  • sarkarsarkar SwitzerlandMember

    Thanks for your reply. I will ask for a newer version. I am just wondering if the number of Actual calls made can be greater than the number of Confidently called bases ? Or this Result is wrong?

  • Geraldine_VdAuweraGeraldine_VdAuwera Cambridge, MAMember, Administrator, Broadie admin

    I'm not sure how those numbers were calculated, or even exactly how they were defined, as we no longer output those metrics.

  • sarkarsarkar SwitzerlandMember

    Thanks for your feedback. I will try with the latest version.
    As per my knowledge, Callable bases that exceed the emit confidence threshold, either for being non-reference or reference and they should be always greater or equal to Actual calls made.

Sign In or Register to comment.