The tximport vignette suggests to just use RSEM’s summary from transcripts to genes (ie, rsem.genes.results.txt) instead of using tximport to do the summary from the transcripts file (i.e., use rsem.isoforms.results.txt, set
type="none", and supplying the appropriate column names as arguments). Playing around I noticed that the main difference between the two approaches is the effective length, which I need to use for differential expression analysis in edgeR. Is there a problem with how tximport gets the effective length from RSEM output? I want to use a different mapping from transcripts to genes and I want to be sure that there is no problem with doing so using tximport.