Marcel Martin

I’ve managed to speed up index generation when indels are allowed by about a factor of three. It’s still about 10 times slower than the (sped up) `--no-indels` version, so...

Thanks for testing! Can you please double check the Cutadapt version you are comparing against? I realized only now that there’s no version 3.8. Version 3.7 was followed by 4.0....

> It's a development version, cutadapt-3.8.dev1+g3852ab0 - presumably what would eventually become 4.0. Ah, that explains it. That particular version is just one commit ahead of 3.7, so it’s essentially...

I benchmarked Cutadapt 3.7-4.3 on Python 3.7-3.11 using data similar in structure to yours and nearly the same command line as you provided, but only 2000 adapters and only 250'000...

I made one last improvement. Just like your data, the data that I generated for the benchmark includes `N` bases (within the part of the read that matches the adapter)....

Maybe it’s time to implement an alternative quality trimming algorithm in Cutadapt. The current algorithm is just a reimplementation of the algorithm in BWA and shouldn’t be changed as that...

I haven’t looked at the window-based methods, but the ERNE-FILTER algorithm from the linked paper seems to work on your example. Here’s a simplified version that doesn’t trim the 5'...

Hi, can you describe how your reads look and which type of adapter you want to remove? How many adapters do you expect per read? If your read looks like...

@yuchen345 Just to clarify, in your example, are `ADAPTER1`, `ADAPTER2` and `ADAPTER3` all the same adapter? The `rightmost` option is only implemented for 5' adapters, but I could make it...

I am not entirely sure about your setup, but please check whether this section in the documentation applies to your situation: https://cutadapt.readthedocs.io/en/stable/recipes.html#trimming-amplicon-primers-from-paired-end-reads To separate trimmed and untrimmed output, you can...