Previously vcfwave was run using a bash script I had kicking around. This made integration into Cactus pretty quick, but wasn't ideal for debugging, since Toil was calling a bash script that in turn was running a bunch of piped commands through parallel. It also didn't support distributed computing.
So this PR re-implements the whole process directly in Toil, which should resolve the above two issues. I'll test it on some large inputs on the cluster before merging (I think the memory resource specs will need some tuning).
Previously
vcfwavewas run using a bash script I had kicking around. This made integration into Cactus pretty quick, but wasn't ideal for debugging, since Toil was calling a bash script that in turn was running a bunch of piped commands throughparallel. It also didn't support distributed computing.So this PR re-implements the whole process directly in Toil, which should resolve the above two issues. I'll test it on some large inputs on the cluster before merging (I think the memory resource specs will need some tuning).