bytedance/Protenix

Toward High-Accuracy Open-Source Biomolecular Structure Prediction.

What it solves

Protenix provides high-accuracy, open-source biomolecular structure prediction. It aims to make advanced computational biology tools accessible and extensible for researchers, offering an alternative to proprietary models like AlphaFold3.

How it works

Protenix uses a foundation model to predict the 3D structures of biomolecules. It supports Multiple Sequence Alignments (MSA) for proteins and RNA, as well as template-based predictions. The system allows for inference-time scaling—increasing the number of sampled candidates to improve accuracy for complex targets like antigen-antibody complexes. It also supports atom-level contact and pocket constraints to incorporate physical priors into the prediction process.

Who it’s for

Computational biologists, structural biologists, and researchers in drug discovery or biomolecular design who need high-precision 3D structure predictions of proteins, RNA, and their interactions.

Highlights

  • High Accuracy: Outperforms AlphaFold3 on diverse benchmarks while maintaining similar model scale and inference budgets.
  • Versatile Model Variants: Includes a model for practical applications (updated data cutoff) and a lightweight "Mini" version to reduce inference costs.
  • Extensible Ecosystem: Part of a suite including PXDesign for de novo protein-binder design and PXMeter for reproducible evaluation.
  • Open Source: Released under the Apache 2.0 License for both academic and commercial use.

Related

  • Project
    aqlaboratory/openfold-3

    An open-source reproduction of AlphaFold3 for predicting the 3D structures of proteins, RNA, DNA, and small molecules.

  • Project
    kyegomez/Open-AF3

    An open-source PyTorch implementation of AlphaFold 3 for predicting the 3D structures of biomolecular interactions, including proteins, nucleic acids, and ligands.

  • Project
    google-deepmind/alphafold3

    An implementation of the AlphaFold 3 inference pipeline for the accurate 3D structure prediction of biomolecular interactions.

  • Project
    lucidrains/alphafold3-pytorch

    A PyTorch implementation of AlphaFold 3 for predicting the 3D structures of biomolecular interactions, including comprehensive tools for PDB dataset curation.

  • Project
    RosettaCommons/foundry

    A tooling and infrastructure framework for protein design, providing a unified interface for generative models like RFdiffusion3 and structure prediction models like RosettaFold3.