Hi, I have a question regarding the normalization process when using tximport for processing Salmon output. Specifically, my workflow involves setting the countsFromAbundance parameter to "lengthScaledTPM" for collating read counts.

My question pertains to the necessity and method of library size normalization after obtaining the txi$counts from tximport. Considering the adjustment made by using "lengthScaledTPM", is further normalization for library size required when using txi$counts for downstream analysis? Additionally, if library size normalization is necessary, does this imply that the output from txi$counts is effectively in Reads Per Kilobase (RPK) format?

I appreciate any insights or recommendations on whether library size normalization is needed in this context and the correct interpretation of the txi$counts output under these conditions.

SG

txi <- tximport(files,
                           type = "salmon",
                           tx2gene = tx2gene,
                           countsFromAbundance = "lengthScaledTPM")

The point of tximport is that it leaves the counts close to their original scale, such that they do still require the offset in statistical modeling to account for sequencing depth. The package handoffs in the vignette all account for this (see tximport vignette for details, or tximeta vignette for the option using Bioconductor's SummarizedExperiment class).

Additionally, if library size normalization is necessary, does this imply that the output from txi$counts is effectively in Reads Per Kilobase (RPK) format?

No. The txi$abundance matrix is in TPM format, if that is what you are looking for.

I appreciate any insights or recommendations on whether library size normalization is needed

Yes, library size correction is needed for the counts, this is intentional / the point of tximport. See vignette for appropriate handoffs to statistical inference (DE) packages.