TrimGalore with 10X data

[golharam]

Hi - I'm curious as to the best way to run Trim Galore on 10X scRNA-Seq data. We have sequenced degraded FFPE samples and are seeing the TSO adapter at the start of read2 and in some reads poly-A tails at the end of read 2. I'd like to trim these reads of both the TSO adapter and the poly-A tail. I tried to use R2 as the only input, but this discarded some of the reads, but the paired R1 is still there. So, I tried to then supply both ends using --Paired, but that's only trimming R1, which I don't want to do. Is there a way to apply TrimGalore to just 1 end, but supplying both ends in case any reads need to be discarded?

[FelixKrueger]

I'm afraid there is currently no way to supply both reads but only trim one of them. The next best way I guess would be to only use Read 2 for trimming with e.g. -a A{20} while using --length 0. This would trim reads for PolyA from the 3' end, but not remove sequences if they get too short. I am not sure how CellRanger deals with very short (or 0 length?) reads, but you might have to use a little script that takes in R1 and R2 at the same time, and apply some length filtering, similar to the step 'validate paired-ends' that Trim Galore carries out internally. Does that help?

[golharam]

It does. I'd like to keep this issue open. I may continue back a PR to enable TrimGalore to work on 10X data.

[golharam]

Looking deeper into this, cutadapt has parameters to handle this, namely -p, -A and -G.

[lcolladotor]

Hi,

We are seeing some of the TSO adapter on read2 as well. From https://teichlab.github.io/scg_lib_structs/methods_html/10xChromium3.html that's AAGCAGTGGTATCAACGCAGAGTACAT. I'm just curious if you could post the full command you used for trimming the TSO adapter from read 2.

Thank you!

Best, Leo

[FelixKrueger]

I suppose this is probably one for @golharam to answer? Maybe you could even post it here for future reference? Best, Felix

[lcolladotor]

For reference, here you can see that sequence on the FastQC report for read2, at the "overrepresented sequences" section

---

Yes Felix, I was asking @golharam =) Thanks for the fast reply!

[golharam]

I don't remember the parameters I used. I ended up using cut adapt directly, but I don't think that solved the issue either. In the end, a newer version of CellRanger trimmed adapters better than any way I tried so we went this using a newer version of CellRanger.