Downsampling step that makes hard links for bams that are already lower coverage than the target triggers re-runs when overlap clipped bai's have a later modification date than the bams #12
Description
CH brought this to my attention. When and if things need to be re-run, if the BAI has a later modification time than the BAM, then the hard-linked files at this step:
trigger re-runs because all the output files have to be later than all the input files, and the BAI's for input are later than the BAMs for output.
I had done the hard linking just to save space, but it is clear that the easy way to fix this is to just cp then bams and the bais (i.e. cp instead of ln there).
In the meantime, touching all the offending bams and bais to have the same modification date (which is before any other downstream products) can get around triggering a re-run because of this.