[0.16.4] - 2026-06-09

Changed

• automatically remove old cache directories

Fixed

• adapted to a change in NCBI metadata (ftp_path now has a trailing slash)

[0.16.3] - 2025-09-30

Changed

• if none of the requested providers are online, raise an error
• added retry for gencode ping (flaky connection)
• increased minimum Python version to 3.9 due to:
  • importlib.resources.files()
  • tar.extractall(..., filter="data")

Removed

• removed norns dependency (which was using the deprecated pkg_resources)

Fixed

• pandas FutureWarning
• unsafe tar.extractall

[0.16.2] - 2025-05-12

Fixed

• Ensembl release versions no longer includes unreleased versions
• unit tests
• upgraded formatters (and fixed the marked grammar & spelling errors)

[0.16.1] - 2023-06-14

Fixed

• fix for NCBI's assembly report header "asm_submitter" instead of "submitter"

[0.16.0] - 2023-05-31

Added

• genomepy search now accepts the --exact flag
• genomepy.Annotation.attributes() returns a list of all attributes from the GTF attributes column.
  • e.g. gene_name, gene_version
  • nice to use with genomepy.Annotation.from_attributes() or genomepy.Annotation.gtf_dict()
• When installing assemblies from older Ensembl release versions, a clearer error message is given if assembly cannot be found:
  • if the release does not exist, options will be given
  • if the assembly does not exist on the release version, all available options are given
  • if the URL to the genome or annotation files is incorrect, the error message stays the same
• new config option: ucsc_mirror, options: eu or us.
  • the mirror should only affect download speed
  • can be nice if the other mirror is down!

Changed

• function get_division is now a class method of EnsemblProvider
• EnsemblProvider class methods get_division and get_version now require an assembly name.
• UCSC data is now downloaded over HTTPS instead of HTTP

Fixed

• genomepy.install() now returns a Genome instance with updated annotation attributes.
• now ignoring ~1600 assemblies from the Ensembl database with incorrect metadata
  • no easy way to retrieve this data

[0.15.0] - 2023-02-28

Added

• you can now tune the cache expiration time in the config
  • create a config with genomepy config generate, then tweak the values as desired.
• support for biopython >=1.80 with pyfaidx update
• raise an informative error when UCSC tools are missing
  • this should only happen in Pip installations

Fixed

• disabling already disabled plugins no longer throws an error
• bgzipping fixes:
  • bgzip works again with python>3.7 (openssl shenanigans. tabix was deprecated for htslib)
  • genome index works with genome install --bgzip (a 2nd is created with the correct naming format)
  • export file works with genome install --bgzip
  • genomepy.install_genome(bgzip=True) returns a Genome class instance with correct paths

[0.14.0] - 2022-08-01

Added

• now using filelock for improved thread safety
• now checking if every API/FTP/HTTP(S) is accessible before proceeding
• genomepy search improvements:
  • text search now accepts regex, and multiple substrings (space separated) are unordered.
  • taxonomy search now returns all hits that start with the given number.

Changed

• switched to pyproject.toml + hatchling for packaging

Fixed

• updated the README and CLI documentation to mention the Local provider

[0.13.1] - 2022-06-21

Changed

• removed unused keys from Ensembl and UCSC databases to reduce their size

Fixed

• added a retry for initializing the diskcache (seq2science/issues/887)
• can now find ensembl urls for genomes not using url_names properly (#205)

[0.13.0] - 2022-06-02

Added

• genomepy search and genomepy genomes can now return the (unfiltered) absolute genome size with argument --size

Changed

• changed caching backend to diskcache (thread safe)
• reduced the local cache size of NCBI (by about half)
  • by only storing assembly summary columns actually used by genomepy

[0.12.0] - 2022-03-28

Added

• genomepy.Annotation.lengths() to retrieve the gene/transcript lengths.
• genomepy.Annotation.from_attributes() can extract any sub-column that pesky attributes column

Changed

• updated Boyle-lab blacklists
• genomepy.Annotation.genes() default changed from bed (commonly containing transcript names) to gtf (gene names)

Fixed

• blacklists now work with GENCODE
• query_mygene no longer filters input.
• genomepy install with local provider now understands you want the annotation if you pass a path to an annotation