This switches VCF export from using whole-genome GBZ to chromosome VG files. This is necessary to plug in the nested VCF support, at least as currently implemented, which requires all paths being indexed.
I don't think it'll affect runtime very much -- maybe on a cluster it could go faster because VCF export now distributed by chromosome. But it could potentially use a lot more memory with --noSplit because you'd be working with a whole-genome vg instead of gbz -- but I think this use case remains pretty obscure.
Right now the interface still requires vcf reference samples be selected as references via the --reference option, but this is no longer necessary since we're not going through GBZ. So it would be a small future change to lift this requirement, enabling VCF creation on any sample in the graph.
This switches VCF export from using whole-genome GBZ to chromosome VG files. This is necessary to plug in the nested VCF support, at least as currently implemented, which requires all paths being indexed.
I don't think it'll affect runtime very much -- maybe on a cluster it could go faster because VCF export now distributed by chromosome. But it could potentially use a lot more memory with
--noSplitbecause you'd be working with a whole-genome vg instead of gbz -- but I think this use case remains pretty obscure.Right now the interface still requires vcf reference samples be selected as references via the
--referenceoption, but this is no longer necessary since we're not going through GBZ. So it would be a small future change to lift this requirement, enabling VCF creation on any sample in the graph.