Skip to content

Latest commit

 

History

History
131 lines (104 loc) · 7.68 KB

CHANGELOG.md

File metadata and controls

131 lines (104 loc) · 7.68 KB

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[Unreleased]

[1.7.0] - 2023-09-13

Changed

  • The mmseqs search command has been replaced by a two-step alignment workflow. In the first alignment step, --alignment-mode 1 and --max-rejected are utilized, while the second step uses --alignment-mode 2 and -c 0.2. This change reduces the number of alignments that are rejected due to not meeting the minimum coverage cutoff and mitigates the issue where the annotation results change when the input sequence order is altered.
  • The --min-ungapped-score parameter of mmseqs prefilter was increased from 20 to 25.
  • The --max-rejected parameter of the first mmseqs align step was increased from 225 to 280.

[1.6.1] - 2023-07-31

Fixed

  • Replace np.warnings with warnings to add compatibility with numpy >= 1.24.

[1.6.0] - 2023-07-31

Changed

  • Update numba (>=0.57) and numpy (>=1.21) version requirements.
  • Use casefold for sequence comparison within the Sequence class.
  • Remove type annotations of methods of the Sequence class that return an instance of Sequence.
  • Use console.status to log the deletion of the .tar.gz file during the execution of download-database.
  • Make the conservative assignment at the family level optional via the --conservative-taxonomy parameter. This increases the amount of viral genomes assigned to a family when executing geNomad with default parameters.

Fixed

  • Fix parameter names in the error message of --conservative and --relaxed (e.g. --min_score → --min-score).

[1.5.2] - 2023-05-11

Added

  • Display a progress bar showing the progress of the classification process in nn-classification.

Changed

  • Update README.md to the database version 1.3.0.

Fixed

  • Make mmseqs convertalis output the whole sequence header instead of gene accesions. This prevents parsing conflits with geNomad's other components in cases where MMseqs2 uses its built-in special parsers for specific header formats (e.g. RefSeq).

[1.5.1] - 2023-03-30

Added

  • Add the --threads parameter to the nn-classification module, which allows controlling the number of threads used for classifying sequences using the neural network model.

Changed

  • Mention post-classification filters the in the summary module description.

[1.5.0] - 2023-03-02

Changed

  • Given that geNomad applies a minimum score filter (since version 1.4.0), the help dialogue of the --min-score parameter was modified to remove the following sentence: "By default, the sequence is classified as virus/plasmid if its virus/plasmid score is higher than its chromosome score, regardless of the value".
  • The following parameters were added to the MMseqs2 search command: --max-seqs 1000000 --min-ungapped-score 20 --max-rejected 225. As a result, changing --splits won't affect the search results anymore.

[1.4.0] - 2023-02-17

Added

  • Mention Docker and the NMDC EDGE implementation in the README.md.
  • Add the --min-plasmid-hallmarks-short-seqs and --min-virus-hallmarks-short-seqs parameters. These options allow filtering out short sequences (less than 2,500 bp) that don't encode a minimum number of hallmark genes. By default, short sequences need to encode at least one hallmark to be classified as a virus or a plasmid.
  • Add the --conservative and --relaxed presets that control post-classification filters. The --conservative option makes those filters even more aggressive, resulting in more restricted sets of plasmid and virus, containing only sequences whose classification is strongly supported. The --relaxed preset disables all post-classification filters.

Changed

  • Windows with more than 4,000 Ns are ignored when encoding sequences for the neural network classification. The first window is always processed, regardless of the amount of Ns.
  • Changed the default value of --min-score from 0.0 to 0.7.
  • Changed the default search sensitivity from 4.0 to 4.2.
  • Update README.md to version 1.4.0. This includes mentions to the --conservative and --relaxed flags and a warning about how changes in --splits can affect geNomad's output.

[1.3.3] - 2023-01-05

Fixed

  • Fix a bug in score-calibration that happened when find-proviruses was executed but no provirus was detected. The module now checks if proviruses were detected (using utils.check_provirus_execution) before counting the total number of sequences.

[1.3.2] - 2022-12-28

Fixed

  • Require numpy <1.6. Fixes #7, which occurs because numba doesn't support numpy >=1.24 yet.

[1.3.1] - 2022-12-22

Fixed

  • Check if find-proviruses was executed when counting the number of sequences in the score-calibration module.

[1.3.0] - 2022-12-12

Added

  • Add support for AMR annotation.
  • Update database parsing to allow BUSCO-based USCGs.

Changed

  • Sequences with no terminal repeats will be flagged with No terminal repeats, as Linear can be misleading.
  • Print the number of plasmids and viruses in the summary module.
  • Set click.rich_click.MAX_WIDTH to None.
  • Reduce the default --sensitivity to 4.0.
  • Update README.md to version 1.3.0.

Fixed

  • Set prog_name in click.version_option.

[1.2.0] - 2022-11-15

Changed

  • Mention the Zenodo upload of geNomad's database in README.md.
  • Add the following sentence for the help dialogue of the --min-plasmid-marker-enrichment, --min-virus-marker-enrichment, --min-plasmid-hallmarks, and --min-virus-hallmarks parameters: "This option will be ignored if the annotation module was not executed".
  • Apply a uniform prior to the empirical sample composition in score_batch_correction. This will shrink the effect of calibration when the empirical composition distribution is very skewed.
  • Reduce the --min-score in the README.md example to 0.7.

Fixed

  • Fix a bug in the score calibration module where the sample size was set to a constant value and the "Your sample has less than 1,000 sequences…" warning would always appear.

[1.1.0] - 2022-08-22

Added

  • Dockerfile for version 1.0.0.
  • Sequence class: add support for str in __eq__.
  • Sequence class: add a __hash__ method.
  • Compute marker enrichment in the marker-classification module.
  • Add columns for plasmid and virus marker enrichment to the _plasmid_summary.tsv and _virus_summary.tsv files.
  • Set --min-plasmid-marker-enrichment and --min-virus-marker-enrichment to 0 as default. This will alter the results when using default parameters.
  • Add support for plasmid and virus hallmarks. Requires geNomad database v1.1.
  • Add CONJscan annotations to _plasmid_summary.tsv. Requires geNomad database v1.1.

Changed

  • Sequence class: simplify has_dtr return statement.
  • Sequence class: make __repr__ more friendly for long sequences.
  • Sequence class: rename the id property to accession.
  • Amino acids are now written to _provirus_aragorn.tsv.
  • Update the XGBoost model file to the .ubj format.
  • Require xgboost >=1.6.
  • The taxonomic lineage in _taxonomy.tsv and _virus_summary.tsv will use Viruses as the highest rank, instead of root.
  • Change order of the columns in _plasmid_summary.tsv and _virus_summary.tsv.
  • Explicitly set fraction to 0.5 in taxopy.find_majority_vote.

Fixed

  • tRNA coordinates are now 1-indexed.
  • Write summary_execution_info.
  • Fix a problem in DatabaseDownloader.get_version where only the major version was compared.

[1.0.0] - 2022-07-12

Added

  • First release.