Hooking up a siHMM (or some variation) to the coverage model from https://github.com/broadinstitute/gatk-protected/issues/1038 will be our first cut at GATK CNV 2.0. Performance and convergence on WGS-size data should be a priority. The end goal will be a joint siHMM once we also have the 2.0 allele-fraction model prototyped.
I think this is obviated by kernel segmentation. We may use some sort of HMM for the tumor heterogeneity tool, but we can open new, more relevant issues then.
@samuelklee commented on Fri May 12 2017
Hooking up a siHMM (or some variation) to the coverage model from https://github.com/broadinstitute/gatk-protected/issues/1038 will be our first cut at GATK CNV 2.0. Performance and convergence on WGS-size data should be a priority. The end goal will be a joint siHMM once we also have the 2.0 allele-fraction model prototyped.