I am trying to count reads of Riboseq data - which ranges from 25 to 37 nt- and RNAseq data, on both gene and transcript level.
For the gene level, I use Featurecounts : I am counting exons for RNA seq and CDSs for Riboseq data ( to count only coding RPFs) , and using their raw counts in downstream analyses to get translational efficiency.
For the transcript level, I was using featurecounts too, by searching exons of transcripts ( instead of genes ) again to have raw counts.
However I recently read on BioStars that it is not recommended and likelihood based counts should be used. A: counting reads with featureCounts - uniquely assign reads to transcripts
I have couple questions regarding:
1) But then, will they be raw counts, in other words, can estimated counts of ML based algorithms qualify for raw counts? (such as suitable enough for Deseq2 to take over the downstream analyses before tximport).
2) Short reads are not very well adopted by maximum likelihood / variational Bayes methods. In this case, with which tool I can use to count my Riboseq data? I wanted to use salmon on transcript and gene level quantification of Riboseq data, but I have read that it is only suitable for processing RNA seq data.
So I am left with no options?
Any suggestions for what to use to count short Riboseq data?