googlegenomics/gcp-variant-transforms

No need to use `reference_name` in merge key

samanvp opened this issue · 0 comments

As we shard variants per chromosome before conducting merge stage, we don't need to use reference_name in the merge key. In other words, since we merge variants of each chromosome independently from other chromosomes, there is no need to use reference_name in the key here:

yield ':'.join([str(x) for x in [variant.reference_name, i]])

and here: