Add hardclip segments option to SplitNCigarReads tool

[jamesemery]

Currently the GATK4 version of SplitNCigarReads softclips the overlapping segments of the reads across the split segments. This is a departure from the original GAKT3 behavior which hardclipped the edges. A few discussions have happened where this has confused users since running HaplotypeCaller/Mutect on the results can often result in confusing indels when they try to align the soft-clipped segments. Currently we can simply tell people to ignore soft-clipped bases in those tools but another solution for users who want to call based on split reads would be to add an option to the tool SplitNCigarReads to call to the hard-clipping machinery instead. We would have to be careful that the mate tags are correctly computed based on the hard-clipping.

[GATKSupportTeam]

(related to Zendesk ticket #140808)

[gbrandt6]

Someone from my team is going to give this issue a try, we'll post any updates here.

[Rmulet]

Are there any updates on this feature? The use of soft-clipping is not only confusing, but can negatively affect the performance of other tools that use this sort of information. Ignoring soft-clipped reads altogether, if possible at all, is not a good solution. We are forced to use GATK3 because the output of the GATK4 version does not work well with others tools we need for the detection of certain variants in RNA-seq.