I should add that I was advised earlier to split it up into much smaller datasets, but colleagues have told me they have managed to run this sizes of datasets. So I am unsure if the size is indeed the problem, but perhaps it is.

I didnt check for that advice, but you could use the "run" column in the samplesheet so that denoising is on smaller datasets. You should use that anyway if the samples originated from several sequencing runs (and I would assume so with that many samples!).)

Ok, so I just add run as a column in the sample sheet? Do I need to adjust flags for that to work? It is one sequencing project, so not sure how to divide it in a suitable way. But I could try e.g. 5 or 10 batches for the denoising and see if that works?

> Ok, so I just add run as a column in the sample sheet? yes, as described in https://nf-co.re/ampliseq/2.16.1/docs/usage/#samplesheet-input > Do I need to adjust flags for that to work? No > It is one sequencing project, so not sure how to divide it in a suitable way. But I could try e.g. 5 or 10 batches for the denoising and see if that works? Yes, thats an idea. > one sequencing project that wording seems ambiguous to me: the question is whether all samples were sequenced in one run of one sequencing machine on the same flow cell. If there were any batches, use those in the "run" column.

Ok, thank you! There were more than one flow cell indeed. I will look into that and do batching.