Counts From Separate Batches

[Liam Salleh]

Hi,

I have two datasets of manually isolated cells, that were done at different times and depths.

I am planning on using tximport with the argument countsFromAbundance="lengthScaledTPM" and using txi$counts as the input matrix (described here) to a SingleCellExperiment object, and then using limma for batch correction.

My question is:

Is there a difference between quantifying the two datasets separately and then reading them all in with tximport, as opposed to running salmon on all of the mate pairs from the two batches at once? (assuming the same reference transcriptome/index file is being used) and is one method more recommended than the other? So far I have run salmon separately for each of my datasets (same index).

Thanks,

Liam