I'm trying to use STAR-Fusion detection. I have RNAseq data that contains UMIs in it. During RNAseq analysis I use them to deduplicate the final bam alignment produced by STAR and continue with downstream analysis, but the output used for Fusion analysis is not the final bam alignment, but the Chimeric.out.junctions, which is not deduplicated. I was thinking about aligning with chimeric reads in the bam file (STAR e --chimOutType WithinBAM), deduplicate the bam file with UMI tools, then filter the chimeric.out.junctions output for the reads I see in the deduplicated bam file.
Is this an ok workaround or is there a simpler way to deal with it? I just don't want the final output of the Fusion analysis to give me inflated junction read counts because of PCR duplication.
