I'm trying to use STAR aligner for ribosome profiling libraries, which have are directional due to the library kit I used (NEB small RNA). But the generally accepted workflow for ribosome profiling read alignment involves eliminating reads which map more than once to the genome (i.e. STAR parameters of --outFilterMismatchNmax 2 --quantMode TranscriptomeSAM GeneCounts --outSAMattributes MD NH --outFilterMultimapNmax 1). As far as I understand it, STAR does not distinguish strandedness in any way other than the geneCounts table output. So, does this that I am losing data by discarding reads which multimap when processed as unstranded, but aren't actually multimappers because they should only be aligned in one direction? If so, does anyone know of a way to get around this issue? Should I expand my multimapping allowance and then do a quantification on the BAM output with a tool that can distinguish read direction (but then how do I make sure the quantification tool only counts the best-scoring read for the proper direction?)?
Similarly, I have been pre-filtering reads against rRNA/tRNA/snRNA with Bowtie prior to STAR alignment, but if I'm not specifying strandedness in that step, could I be losing a lot of reads that might actually be legitimate mRNA-mapping reads? If anyone has suggestions for alternative tools that allow for specifying directionality, that would be greatly appreciated.