We’re moving the GATK website, docs and forum to a new platform. Read the full story and breakdown of key changes on this blog.
If you happen to see a question you know the answer to, please do chime in and help your fellow community members. We encourage our fourm members to be more involved, jump in and help out your fellow researchers with their questions. GATK forum is a community forum and helping each other with using GATK tools and research is the cornerstone of our success as a genomics research community.We appreciate your help!
Test-drive the GATK tools and Best Practices pipelines on Terra
Check out this blog post to learn how you can get started with GATK and try out the pipelines in preconfigured workspaces (with a user-friendly interface!) without having to install anything.
ReadBackedPhasing --respectPhaseInput

The 2013 "best practices" workshop slides recommend running PhaseByTransmission followed by ReadBackedPhasing --respectPhaseInput.
The --respectPhaseInput option is not currently listed in the documentation. Does that mean that RBP now always respects phasing in the input VCF?
Does (or did) --respectPhaseInput cause phased sites in the input to be assumed correct, or are they just ignored? That is, does RBP --respectPhaseInput use the partial haplotypes from the input file as evidence?
Thanks!
Douglas
Best Answer
-
Laurent ✭✭
Hi @dmyersturnbull,
PhaseByTransmission operates on a site-by-site basis, so it will be totally blind to previous phasing annotation. You can of course phase sites with PBT and then merge them with the previous phasing, but this will require a little file manipulation. Note that if you're willing to get your hands dirty, it would be very easy to add a line in the PBT code so that it doesn't process previously-phased sites. Although it would be best for Menachem to respond, I believe that when using RBP with the --respectPhaseInput option will indeed only phase sites that weren't previously phased and do so while respecting previous annotations.
Cheers,
Laurent
Answers
@dmyersturnbull
Hi Douglas,
You no longer need to use ReadBackedPhasing. HaplotypeCaller does physical phasing for you. We only use ReadBackedPhasing for merging MNPs. The updated slides from our latest workshop are here: https://www.broadinstitute.org/gatk/blog?id=5338
It is true the --respectPhaseInput argument is gone. We will check with the developers on what the current behavior is for dealing with phased input.
-Sheila
Hi Sheila,
I'd forgotten that about HaplotypeCaller.
Will PhaseByTransmission respect previous phasing or use it as evidence? I noticed that HaplotypeCaller uses the annotations PID and PGT rather than GT (I never see a | in the GT from HC). Does PBT use those?
I'm trying to use ShapeIt2 for phasing, but it can't phase variants that aren't in the reference panel (1000G phase 3), so I'm hoping to fill those in afterward using PBT and either RBP or HC's PGT annotations. Filling in the unphased GTs using PGT annotations seems like it could result in inconsistent haplotypes, so I'm wondering whether PBT and RBP can "build off of" ShapeIt2's partial phasing. That's why I'm interested in this.
Thanks,
Douglas
@dmyersturnbull PhaseByTransmission doesn't know about HC's phasing tags, so it will neither take them into account nor change them in any way. The fact that they store the phasing information in different forms is done on purpose to avoid collisions.
I looked briefly at the PBT and RBP code but it wasn't obvious to me how the tools treats incoming phasing info. The phasing tools are developed by external collaborators, @mfromer and @Laurent -- hopefully they can have a look and answer your questions. If not I'll try to hunt down the answer from the rest of the team next week.
Hi @dmyersturnbull,
PhaseByTransmission operates on a site-by-site basis, so it will be totally blind to previous phasing annotation. You can of course phase sites with PBT and then merge them with the previous phasing, but this will require a little file manipulation. Note that if you're willing to get your hands dirty, it would be very easy to add a line in the PBT code so that it doesn't process previously-phased sites. Although it would be best for Menachem to respond, I believe that when using RBP with the --respectPhaseInput option will indeed only phase sites that weren't previously phased and do so while respecting previous annotations.
Cheers,
Laurent
Thanks, @Sheila, @Geraldine_VdAuwera, and @Laurent. That clears it up completely!
Best,
Douglas