What's changed

• Fix for #570
• Updated packaging metadata and build configuration (new pyproject.toml, adjusted setup, and environment changes), including adding the package data module for bundled assets.
• Modernized Snakemake integration, allowing update of dependency. This removes the direct call to the snakemake API, which is deprecated in recent versions and instead handles Snakemake calls in a new function in misc (imaginatively named dir).
• This removed the legacy custom logger utility, as it is no longer compatible with the updated api calls, consequences of this will be more verbose output to screen from snakemake as snakemake is running within pangolin, but this is a small change and allows us to update a lot of the pins so worth it.
• Refactored utilities around I/O and initialization to support the new Snakemake call path.
• Adjusted dependency checks and data validation logic to align with the updated packaging approach.
• Minor metadata update in pyproject (author info changed to cov-lineages team).

What's Changed

• Add entries for pangolin-data/-assignment 1.36 by @AngieHinrichs in #566
• Fix Docker image by @AngieHinrichs in #568

Full Changelog: v4.3.3...v4.3.4

What's Changed

• conda channels: change defaults to nodefaults by @AngieHinrichs in #563 HT @ammaraziz
• Dockerfile: use mamba, update alpine image by @AngieHinrichs in #565 HT @abel-betraoui

Full Changelog: v4.3.2...v4.3.3

Minor update: environment.yml constrains usher to <=0.6.3 because problems with 0.6.4-0.6.6 have been reported (#560)

Full Changelog: v4.3.1...v4.3.2

Patch update to resolve snakemake dependency issue #525. Required update to env file, pinning of snakemake-minimal version to 7.24.0 and update to github actions to pass tests. Later versions of snakemake-minimal would not solve in github actions.

• pangoLEARN analysis mode is now deprecated due to lack of resources for updating the pangoLEARN model. If --analysis-mode fast or --analysis-mode pangolearn is given, pangolin v4.3 will print out a warning and use UShER mode instead, unless --datadir is also given specifying a directory with pangoLEARN model files. The next release of pangolin-data (v1.20) will no longer include the model files which have not been updated since v1.18). (PR #519)

• In UShER mode, in the case of multiple equally parsimony-optimal placements (EPPs), the lineage with the most EPPs will no longer override UShER's selected lineage. This "voting" logic was problematic in the case of amplicon dropout (e.g. issue #492). (PR #521)

• dependencies: snakemake-minimal is no longer pinned to <=6.8.0 and tabulate is no longer pinned to 0.8.10. (PR #514)

Significant speed improvement for the default UShER analysis mode when running on a large number of input sequences, if the latest version of usher (v0.6.1 or later) is installed (PR #499: Use usher-sampled if installed).

Patch to pin tabulate version described in issue #489.

• merging PR #474
• --all-versions flag now also has faToVcf version

Patch to fix --all-versions flag. Everything bar faToVcf now reported, will need faToVcf to have --version flag for this approach to be compatible.