Some processes are short running but numerous like mafft such that most of the time spent running on HPCs and clouds is waiting got the job scheduler. We could batch these jobs based on the first letter in a hash, meaning an extra job would cause just one batch to be rerun. Or some other grouping, like one mafft job for all genes for a tree group.
Some processes are short running but numerous like mafft such that most of the time spent running on HPCs and clouds is waiting got the job scheduler. We could batch these jobs based on the first letter in a hash, meaning an extra job would cause just one batch to be rerun. Or some other grouping, like one mafft job for all genes for a tree group.