Good afternoon, I am trying to run KneadData (version 0.12.0) on the iHMP IBD MTX dataset and on an IBS dataset. I noticed that paired reads that do not have the same read length end up in their respective unmatched fastq files. Hence, I end up with almost empty paired fastq files. As I would like to assemble contigs using rnaSPAdes, I would need more paired reads given that with the current KneadData output, rnaSPAdes fails for most samples. Is there a way to force KneadData to write unmatched paired reads into paired fastq files even if paired reads do not have the same read length?
By further inspecting the raw reads fastq files and the KneadData outputs, I concluded that the read length is encoded in the reads’ headers, causing paired reads with different reads lengths to be considered as unmatched reads by KneadData. The solution would be to remove the reads lengths from the reads headers.