I downloaded the SILVA ribosomal RNA database from your website (downloaded directly because kneaddata_database failed for me), and I have a question. Looking at the index with bowtie2-inspect, it looks like the reference sequences contain only A/C/G and that the U’s present in the SILVA fasta files have been removed. Building a custom database using bowtie2-build with the downloaded FASTA files has the same problem (is it best to just convert the U’s to T’s in the SILVA fasta file?). My apologies if I’m missing something obvious - this is my first analysis with kneaddata, and my first analysis where I actually care about characterizing rRNA in an experiment.
In any case, thanks for developing this very useful tool!
PS - I actually posted this question at https://bitbucket.org/biobakery/kneaddata/issues, but I assume this is a better spot…