Skip to content

v0.5.3: Protenix constraint/mini models release

Choose a tag to compare

@zhangyuxuann zhangyuxuann released this 17 Jul 03:58
· 122 commits to main since this release

What's Changed

  • Several smaller and more efficient models have been released alongside the main checkpoint. Users can choose the appropriate variant based on their hardware and performance requirements.
  • Constraint feature of Protenix is released! Added atom-level contact and pocket constraints, significantly improving performance in our evaluations.
  • Model inference command changes
    • The model_name is "protenix_base_default_v0.5.0" by default, you can modify it by passing --model_name ${model_name}
    • Each model has recommended default configuration parameters. To customize cycle/step/use_msa settings, set --use_default_params false
    • You can modify cycle/step/use_msa by passing --cycle x1 --step x2 --use_msa false
  • Other minor changes
    • Image was upgraded to v0.0.3, including ESM-related dependencies..
    • Added a progress bar to tell users the download progress of the current model weight and data for inference.
    • Added github "publish to pypi" workflow and Added long_description for protenix package in Pypi.
    • Added link to Protenix evaluation tool PXMeter.
  • Throw error on invalid msa type by @emmanuel-ferdman in #143
  • docs: Grammar Fixes in pairformer.py by @OccupyMars2025 in #147
  • docs: fix typos by @OccupyMars2025 in #150
  • just trivial modification: Update gen_ccd_cache.py by @OccupyMars2025 in #152

New Contributors

Full Changelog: v0.5.2...v0.5.3