STAR to remove polyA tails in QuantSeq 3' data - nf-core/rnaseq pipeline

[MiriamAng]

Hi Alex,
I have the need to remove polyA tails from raw reads coming from QuantSeq 3' data that I am processing using the nf-core/rnaseq pipeline. I came across the Slack nf-core/rnaseq channel where it was mentioned the possibility to perform polyA removal using the STAR parameter --clip3pAdapterSeq, which for my purposes would be a great solution since with Trimgalore I cannot remove both adapter and polyA. I therefore run the pipeline using Trimgalore to perform adapter and quality trimming and I leave STAR taking care of polyA. However it is not clear to me how the length of the As specified influences the trimming. Leaving aside the clip3pAdapterMMp parameter, if I set --clip3pAdapterSeq AAAAAAAA, does it mean that it will remove only a number of As equal to the one specified? If the actual number in the raw read is lower or higher than the number specified, what is the behavior? I don't know a priori how many As will be in my reads, the number is highly variable, therefore I was wondering whether this parameter can be used also in my case.
Thank you for your time!
Miriam

[alexdobin]

Hi Miriam,

In your example, --clip3pAdapterSeq AAAAAAAA will search for 8A sequences throughout the read, with mismatches according to the --clip3pAdapterMMp. It will remove the leftmost sequence and everything after it.
At the end of the read, it will remove any A-polymer, even 1A.
This aggressive trimming may remove "real" (genomic) poly-A in the middle of a read.
I would recommend supplying the sequence that is as long as the actual adapter sequence to avoid overtrimming.