Hi Gordon, in the absence of any description of catchSalmon, I have inferred from the code precisely what it does. Would you be willing to provide an explanation (1) for the definition of the overdispersion coefficient (sum of squares over row mean), (2) for the prior (use of qf and median) and (3) why it is ok to input the counts scaled with the transcript-wise overdispersion coefficients into edgeR, voom etc.. Thank you. Comments from others to whom this is obvious are also much appreciated.

See catchSalmon for tx-level DE

This forum is to provide help with the software syntax and I can help you with that if you have any questions about it. I prefer however to use the refereed literature for publishing mathematical justifications or comparative studies. I understand you might not want to use catchSalmon until it is supported by a published paper, and that's fine.