When I use the q2-cutadapt plugin I realized that it removed the majority of the primers as expected, however it writes all the reads by default even if they didn’t have the primers. I saw the standalone software had this option
Is it possible to add the “–discard-untrimmed” command of the standalone cutadapt software to q2-cutadapt in order to remove reads that don’t have the adapters?
I would like to express my intrest in this topic. Currently I am not using qiime2 cutadapt but the original version, just because of that feature.
I cant see a reason to not discard untrimmed reads. I end up discarding usually ~3% read pairs (occasional low quality samples up to 20%) but feel confident about the remaining sequences.
Why do you think primer-untrimmed sequences are worth keeping (aside from library preparation that provides immediately primer free reads)? What do you think reads without primer sequence could be?
Are you going to expose the “discard untrimmed” argument for cutadapt eventually?
Hey there @DaS!
It is a technical reason — QIIME 2 currently doesn’t allow for “empty”
SampleData[PairedEndSequencesWithQuality], which could happen in the case where no reads are discarded. This is entirely an issue within QIIME 2 — there are a few ways to solve it, including:
- error if no reads are discarded / discarded reads file is empty
- update the existing format definitions to allow for emptiness
- change the framework to support the notion of a “nullable” type
We currently have no ETA for this, but, contributions are always welcome!
Hi, I was just wondering if there had been any progress on the discard untrimmed reads option? This would be so useful to me as I’m using different amplicons with the same barcodes. At the moment I need to use other pipelines, and it would be great to stick with this one.
No — the issue above is still open:
You can track that issue to stay updated.
As a workaround for the time being, you could use an alignment search tool external to QIIME 2 to filter out sequences that contain a specific primer, and them import to QIIME 2. I am not sure if VSEARCH can perform alignment on fastq data, but that would be a good place to start.