In the October 2019 Bioconductor release, we included some new functionality in tximport (so v1.4.0) to import alevin scRNA-seq inferential replicates. We just noticed there was a bug in the import code such that the cells in the inferential replicate matrices (e.g. columns in matrices in
txi$variance) were not in the correct order with respect to
txi$counts, though they had the same ordering of cell names. The
txi$counts cell names (barcodes) were correct, while the variance and infReps matrices were mislabeled. Likely not too many groups were computing or importing these inferential replicate matrices, but wanted to make sure it was known to update tximport if you were.
This is fixed in tximport v1.4.2 (release branch), and in devel branch, so the matrices will be in the same order and with correct cell names (colnames).
Another thing we added just now to devel branch was the ability to filter the imported cells to those in