Losing unmapped reads after alignment

[anoronh4]

STAR_ALIGN process is not keeping the unmapped reads in the BAM. as a result, kallisto is not attempting to map any reads that STAR was unable to map and we may be losing some of the advantages of kallisto. in addition, when a sample does not have UMI the reads that STAR is not able to map are included in kallisto, which means the two workflows are somewhat inconsistent.

umitools dedup does not handle read pairs where r1 is unmapped, so one way to address this issue is to set aside the unmapped reads in a separate set of fastqs (which STAR can probably do) and add them to Kallisto's input.