And Case 2) The reason why we spotted this is mainly because

sortMeRNA

only identified and removed only 3% of reads via the

silva-euk-18s-id95.fasta

and

silva-euk-28s-id98.fasta

files. This low amount of rRNA got me suspicious, so I did a manual test of the same file in a quick alignment with Bowtie2 (command was

bowtie2 -x rRNA_silva_euk -U file

), which identified a

51.61% overall alignment rate

So the question is: Why does sortMeRNA not detect this, and then a consequence, why does STAR not report this as mult-mapping but as

unmapped other

? I have tested it

sortmerna --ref rDNA.fa --reads file.fastq.gz --aligned rRNA_reads --fastx --other non_rRNA_reads --threads 32  -v --index 0 --workdir .

and also with

--no-best

as had been suggested earlier, but the results were pretty much identical using a different rRNA.fa file (>KY962518.1).

Results:
    Total reads passing E-value threshold = 21172086 (14.88)
    Total reads failing E-value threshold = 121075196 (85.12)