we could also remove the raw samples. Not the original ones, but kneaddata writes the original fastq again in the out dir (I think by default) so that's redundant with the input files. We only really need the paired and unmatched fastq to proceed.
Also if we need to save space I think they could be .gz as most software that takes these in can handle gz as input
here : https://github.com/jflucier/ILL_pipelines/blob/102066e9be28d84a18c77ff7f84a3583277b272a/scripts/preprocess.kneaddata.sh#L178
we could also remove the raw samples. Not the original ones, but kneaddata writes the original fastq again in the out dir (I think by default) so that's redundant with the input files. We only really need the paired and unmatched fastq to proceed.
Also if we need to save space I think they could be .gz as most software that takes these in can handle gz as input