Hi team, I am seeking advice on how is the best solution to the problem we are having in smrnaseq with fastp module. There are protocols that will need to trim N nucleotides at the end of the read. Fastp has a parameter for that, but it does that before adapter trimming, so it is not useful. I don't know if we could add another module for this specific case, or somehow adapt the script in fasp to do a second round of trimming when this parameter is on (

trim_tail1

) . Normally the parameter would be set up with

task.ext.args

..., if somebody has some elegant way to do this, let me know.

I'm pretty sure that cutadapt has the option to do this

yes, with cutadapt we could do NNNN, but It seems with fastp is not an option, I can talk to smrnaseq devs to see if it is worth to go back to cutadapt, but everybody was supporting the idea of fastp

Is it worth asking the fastp devs?

there is an issue open as well

If you need to make a local module that calls fastp twice, that is fine

ok, i am doing the local module for now, thanks

Might be nicer to patch if you're just appending a string to the command? Then you'll still be able to update when new versions are available etc, rather than having to remember to update yourself manually.

it won't be just a string though, i will need to add code to run twice fastp, and there is already three type of running there

Fair enough 👍🏻 oof yeah it's pretty complex already

You can call the same module multiple times using aliases, though? No need to make separate implementations for it, I think?

True, just run the module twice.. Would mean a larger intermediate file footprint though, as can't stream

BBDuk would have a

tp=

parameter:

(trimpad) Trim this much extra around matching kmers.

, so you could trim the adapters plus N bases in one go.

I tried to run the module twice, but it is working with symlinks and it was not working, or I couldn't make it work

One thing that came to my mind…can’t you instead of changing the tool or its arguments create a modified

adapter.fasta

adding the additional bases to the adapter sequences?

4^4 - 256 adapters. Not sure if that would slow down the trimming, looking for so many sequences? 🤔

I think that is too pessimistic: The tool may outright support adapter sequences with Ns and allow for mismatches (also because of sequencing errors). I think it depends on the kmer index length how many more kmer sequences you will actually create from those highly similar references.