Open xiangpin opened 1 year ago
@mikelove in case you have any opinion
Great!
Thanks for the PR @xiangpin !
Maybe I'm missing the obvious but why should we add a assay that is a dataframe where DESeq2 needs counts. I fail to see the use-case could you be more explicit?
I would say that having a data.frame instead of matrix in the SummarizedExperiment is not typical.
Yes, a data.frame
in the SummarizedExperiment
is not typical. But for many users, they might also import a data frame to assays
when using the SummarizedExperiment
function to build a SummarizedExperiment
object, since the help information of SummarizedExperiment
show that assays: A ‘list’ or ‘SimpleList’ of matrix-like elements, or a matrix-like object (e.g. an ordinary matrix, a data frame, a DataFrame object from the ‘S4Vectors’ package, a sparseMatrix derivative from the ‘Matrix’ package, a DelayedMatrix object from the ‘DelayedArray’ package, etc...).
and data frame might be more similar to other objects for many users.
As the original SE is not changed (but only just the DESeq2 object, passed to the internals for reproducibility, is), I think this PR is great.
@xiangpin The only thing I ask before pushing is a unit test that would fail before PR (current master) and produce the expected result with the PR.
Thanks for this PR!
The related issue
when convert it to matrix, it works