Does anyone have any experience using `samtools ma...
# modules
m
Does anyone have any experience using
samtools markdup
wrt to marking duplicates with UMI’s?
cc @Simon Pearce, this might also interest you cfr the UMI options of Picard
👍 1
m
Not with
samtools markdup
, but with some other tools if that helps.
m
Do you have any suggestions? Currently we have UMI’s in the read name (which samtools can handle), but I’m looking for potential pitfalls to avoid.
m
In #rnaseq , umi-tools has been the default for several years already, umicollapse was a recent addition, so for both nf-core modules exist. If I do only want to mark reads and not remove them, I am using rumidup mostly.
👀 1
m
What seems to be the advantage of using
rumidup
(love the name btw) over other tools?
it seems to be very similar the the samtools approach
m
Haven’t done an extensive comparison, just a weird soft spot for all tools written in Rust. 🤪 > it seems to be very similar the the samtools approach Rumidup uses a Burkhard-Keller Tree internally, I do not know what data structure/algorithm samtools uses. I was mainly looking for a tool that allows for single mismatches in the UMI sequence due to sequencing errors, because my first dataset had a crappy sequencing quality for the UMI part and then stuck with it due to convenience.
m
Thanks for the explanation! 😄
m
I just recalled that there is of course also
fgbio
- which powers #fastquorum : https://fulcrumgenomics.github.io/fgbio/tools/latest/GroupReadsByUmi.html
s
Yes, although fgbio does consensus building rather than deduplication