porechop_abi on oxford nanopore hq (Q20+, Q30+)

[Ales-ibt]

Hello there! Thank you for developing this tool, it is amazing. As it finds adapter sequences de novo in the raw-reads, it should work as well with the new generation of Oxford Nanopore high-quality reads, right? Do you think any parameters should be adjusted for this task?

Thanks in advance.

Alejandra.

[qbonenfant]

Hi,
It is always nice to see our tool appreciated, so thank you.

To be honest, I can not confirm with full confidence that Porechop_ABI will work better on these new reads, as I do not know what (or even if) ONT changed on the adapter side ( I may need a little update on this matter).
If they kept them 'as is ', yes it should be better.

On the parameter side, it would be hard to give a definitive answer without a proper benchmark (I will put that on my TODO list). From experience, i would mainly recommand you increase the k-mer size around 20 / 22 (sound like a reasonable value), maybe higher if you have really good reads.

This should increase the quality and overall precision of the final adapter, as you will allow less potential variation around the expected adapter sequence. (same number of error allowed in each k-mer, but spread out more).

To change the k-mer size, copy the ab_initio.config file, change the 'k' value, and use the '-abc' option with your custom configuration file path.

I would highly recommand you check the resulting adapters before trimming, as this is the longest step.
What you should look for is consistent adapter inference accrosse multiple runs (use options -go to only print the found adapter sequences instead of trimming the whole file each time).

If you go that route, I would be interested with the result.