Identify RNA-seq reads containing polyA sequence

Question

[goodez]

This may seem like a weird question, but we need to filter our RNAseq data for reads that contain polyA. The data is stranded RNA-seq, 50 bp reads. Would it be easier to find these reads before or after alignment? To be clear, the 50 bp read needs to contain a stretch of polyA, not just come from a transcript containing polyA. Has anyone done this type of analysis?

Answer 1

[GenoMax]

Use bbduk.sh from BBMap suite in filter mode (don't specify ktrim= or qtrim= options) with literal=AAAAA to filter the reads out (adjust length of A's as needed). Use with original data.

[goodez]

Thanks! I have some additional questions then. Since it is stranded RNA-seq, the polyA will actually be stretches of TTTTTT right?

Also I used grep to look for reads containing this, and many of the TTTTTT stretches are in the middle of a read. It doesn't seem possible that the polyA could be surrounded by other sequence on both ends.

[GenoMax]

If you are capturing second strand then yes. Past the TTTTT the sequence may be going into adapters. You can easily check that by trimming reads you filter and select.

Answer 2

[Buffo]

If data is stranded, polyA tails will be always at the end of sequences, I have read a couple of papers where this information is important (to define UTRs mainly, if I remember the reference I will post it). They usually consider a 8-10 nucleotides as the minimum length.

[goodez]

Thanks, that is good to know. Please do share the reference if you find it again!