Are UMI reads not sequenced more than once removed during dedup step?

[goyastephanie]

Hi! I would like to confirm a detail about the UMI deduplication process: If there is a single read with a given UMI without any copy of it (due to different reasons, such as not enough sequencing depth), after the UMI dedup step, is this read kept in the deduplicated output? Or is the read filtered because no deduplication/correction steps could be performed on this UMI?

[IanSudbery]

UMI tool's approach is that it removes things it definately thinks are deuplicates. As such, where there is only a single read at a given coordinate, that read cannot be a duplicate, and so the read is retained.