Long colnames get truncated

[FelixTheStudent]

Writing out a sparse Matrix seems to work fine (no warnings). However, once I read it back it, colnames get cropped so that they are not longer than 11 characters (in the minimal example below at least; I have had another case where colnames were not longer than 10).

I use HDF5Array 1.12.3.

library(Matrix)
library(HDF5Array)

# create example matrix with increasingly long colnames:
mat <- as(matrix(rep(0, 900), nrow=30), "dgCMatrix")
mat[sample(1:30, 3), sample(1:30, 3)] <- 1
colnames(mat) <- sapply(1:30, function(i) paste(rep("x", i), collapse = ""))
# write out:
HDF5Array::writeTENxMatrix(mat,
                filepath=file.path("~", "tmp_file.hdf5"),
                group="tmp", level=NULL, verbose=F)
x <- HDF5Array::TENxMatrix("~/tmp_file.hdf5", group="tmp")
max(stringr::str_length(colnames(mat)))
# 30
max(stringr::str_length(colnames(x)))
# 11

Any ideas?

[hpages]

Thanks Felix. Should be fixed in HDF5Array 1.15.7 (see commit 0e487076). Will port the fix to BioC 3.10 later today.

H.

[FelixTheStudent]

Excellent, thanks. Will tell my admin to update bioconductor.

Cheers,

Felix

Hervé Pagès notifications@github.com schrieb am Fr., 28. Feb. 2020, 12:06:

Thanks Felix. Should be fixed in HDF5Array 1.15.7 (see commit 0e48707 https://github.com/Bioconductor/HDF5Array/commit/0e487076bb7faf028a5fd54bd6c7579ce12b268c). Will port the fix to BioC 3.10 later today.

H.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Bioconductor/HDF5Array/issues/24?email_source=notifications&email_token=ACQASS2A5IAVCEVQR53GXJ3RFDV4HA5CNFSM4K5M3AX2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOENIE56A#issuecomment-592465656, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACQASS4LPPAYN4HLQL6L75LRFDV4HANCNFSM4K5M3AXQ .

[hpages]

Sounds good. I've ported the fix to BioC 3.10. It's in HDF5Array 1.14.3 but please allow 24h for the package to pass the builds and become available via BiocManager::install().

Note that admins should update Bioconductor on a regular basis regardless (dozens of packages get fixed every month in release).

Cheers, H.