question on UMI deduplication step

[sparthib]

Hi there,

In my output I see, matched_reads_dedup.fastq and align2genome.bam . My question is, does UMI based deduplication occur at the fastq level, and is this dedup.fastq file then used for alignment? I just want to make sure that my BAM file doesn't have duplicate UMIs.

Thanks, Sowmya

[ChangqingW]

Our current UMI dedup is rather simplistic, it will just keep the longest read with the same UMI, rather than doing any consensus calling. The deduped one is used when realigning to the transcriptome (but not the initial align2genome.bam), you can double check this with samtools view -H align2genome.bam and the last few lines should tell you which command was used to produce the BAM file.

[sparthib]

Thanks, so if I understand correctly, the genome alignment still contains the duplicate reads but the transcriptome alignment doesn't?

[ChangqingW]

Yes, that's correct