This is mostly for cactus-pangenome which, by chaining together a bunch of workflows, has plenty of opportunity to accumulate jobstore bloat. This PR adds some following cleanup steps
erase fasta and paf after consolidated (just paf in progressive mode due to outgroups)
original fasta and minigraph output after graphmap-split
chromosome hal_ids after merging together
On the yeast graph, this drops peak usage from 1.1G to 896G. I think it'll have more impact on big human graphs where sequence data accounts for a bigger proportion of disk usuage.
This is mostly for
cactus-pangenomewhich, by chaining together a bunch of workflows, has plenty of opportunity to accumulate jobstore bloat. This PR adds some following cleanup stepsgraphmap-splitOn the yeast graph, this drops peak usage from
1.1Gto896G. I think it'll have more impact on big human graphs where sequence data accounts for a bigger proportion of disk usuage.