Changes in version 0.2.7                        

  - NanoString nCounter .RCC files are read as counts
  - Count tables whose sample columns are numeric ids (key 2683 2685
    ...) are read with that header instead of as headerless data.
  - Downloads retry HTTP 408, 429 and 5xx (NCBI answers 503 under load)
    and discard the error page before retrying.

                        Changes in version 0.2.6                        

  - download_supplementary(), download_runs(), download_bams() and
    download_dump() now require dest_dir. They used to default to a
    directory under the working directory; name it explicitly, e.g.
    DownloadRuns("SRR12012336", "SRR12012336").
  - New seqout_online() (SeqoutOnline()) reports whether the API is
    reachable, checked once per session.
  - seqout_matrix() reads .h5ad and 10x .h5 files that h5py compressed
    with LZF (GSE300265) when Bioconductor's rhdf5filters is installed

                        Changes in version 0.2.5                        

  - seqout_matrix() no longer aborts with "'length = N' in coercion to
    'logical(1)'" on tar and .rds units (GSE192693, GSE255298). The
    units table's per-unit assay column was shadowing the requested
    assay inside the reader.
  - genevintage moved from Imports to Suggests. SeqoutGeneNames(),
    SeqoutCorrectNames() and SeqoutSex() ask you to install it from
    https://saketlab.r-universe.dev when it is missing.