Latest GATK and DINDEL

[adaptivegenome]

The latest version of GATK does not appear to have a DINDEL option for the indel caller. Is DINDEL the default method for indel calling with the GATK Unified Genotyper?

Has anyone compared the two tools to confirm this?

[RDW]

The latest version of GATK does not appear to have a DINDEL option for the indel caller. Is DINDEL the default method for indel calling with the GATK Unified Genotyper?

Has anyone compared the two tools to confirm this?

The current method is called INDEL rather than DINDEL, and it's not the same. See this reply to a similar query from Eric Banks at GSA:

http://getsatisfaction.com/gsa/topics/_glm_dindel_is_not_working_in_gatk_version_1_0_5777

"In our own tests the newest model performs best, which is why we transitioned over to it. To be honest, though, the genotype likelihoods model for indels is still in constant flux and we are continuously improving it...The reason why we renamed it, by the way, is that this newer model is not really related to the actual "Dindel" software while the previous one was very much inspired by it. We felt that it wasn't fair to the authors of Dindel to continue naming it after their tool."

[adaptivegenome]

The current method is called INDEL rather than DINDEL, and it's not the same. See this reply to a similar query from Eric Banks at GSA:

http://getsatisfaction.com/gsa/topics/_glm_dindel_is_not_working_in_gatk_version_1_0_5777

"In our own tests the newest model performs best, which is why we transitioned over to it. To be honest, though, the genotype likelihoods model for indels is still in constant flux and we are continuously improving it...The reason why we renamed it, by the way, is that this newer model is not really related to the actual "Dindel" software while the previous one was very much inspired by it. We felt that it wasn't fair to the authors of Dindel to continue naming it after their tool."

I appreciate this. So it seems the method is similar or the same, but Eric and his group are continuously updating the error model? So I think I will have to compare both to see how they perform on my data.

Thanks again for the post.

[RDW]

So it seems the method is similar or the same, but Eric and his group are continuously updating the error model?

Just to clarify, my understanding is that the DINDEL option they were using for the Unified Genotyper a few months ago was indeed similar to the freestanding Dindel program from Sanger, but the INDEL option they're now using is based on a different model (which continues to evolve). So all the more reason to do the comparison you suggest (and post the results!).

[adaptivegenome]

Just to clarify, my understanding is that the DINDEL option they were using for the Unified Genotyper a few months ago was indeed similar to the freestanding Dindel program from Sanger, but the INDEL option they're now using is based on a different model (which continues to evolve). So all the more reason to do the comparison you suggest (and post the results!).

Yes, I will post what I find...