steineggerlab/foldseek

Foldseek enables fast and sensitive comparisons of large structure sets.

What it solves

Foldseek is designed for the fast and sensitive comparison of large protein structure sets. It addresses the challenge of searching and clustering protein structures at scale, allowing researchers to find similar structures in massive databases like AlphaFoldDB and PDB in seconds.

How it works

Foldseek uses a structural alignment approach to compare proteins. It supports several alignment modes, including a default local 3Di+AA alignment, global TMalign, and local LoLalign. It can operate on CPU or GPU for acceleration. A unique feature is its ability to to use a language model (ProstT5) to perform comparisons directly from protein sequences, bypassing the need for predicted structures.

Who it’s for

This tool is primarily for structural biologists and bioinformatics researchers who need to perform large-scale protein structure searches, alignments, and clustering.

Highlights

  • High Performance: Supports GPU acceleration for faster searches and clustering.
  • Versatile Input: Accepts protein structures (PDB/mmCIF) or protein sequences (FASTA) via a language model.
  • Comprehensive Search: Supports both monomer and multimer searches and clustering.
  • Integrated Visualization: Includes a built-in terminal structure viewer called StrucTTY.
  • Pre-built Databases: Provides easy access to pre-generated databases such as AlphaFoldDB and PDB.

Related

  • Project
    ccsb-scripps/AutoDock-GPU

    An OpenCL and CUDA accelerated version of AutoDock 4.2.6 that leverages GPUs to speed up molecular docking and virtual screenings of ligand-receptor poses.

  • Project
    sokrypton/ColabFold

    A tool that makes protein folding predictions accessible via Google Colab and local installations, integrating multiple models like AlphaFold2 and AlphaFold3 with a public MSA server.

  • Project
    rdk/p2rank

    A machine learning-based command-line tool for fast and accurate prediction of ligand-binding sites from protein structures.

  • Dispatch
    Deep Learning with Proteins – Hugging Face guide to protein language models and folding

    Hugging Face announced a tutorial series showing how to fine‑tune protein language models and use ESMFold for protein folding, demonstrating that transfer learning techniques from NLP can be applied directly to protein sequence tasks.

  • Project
    aurekaresearch/OpenDDE

    An all-atom biomolecular foundation model that uses co-folding for structure prediction, design, and optimization in drug discovery.