Skip to content

Features Database

Jump to bottom
Dima Molodenskiy edited this page Aug 21, 2026 · 3 revisions

Instead of generating feature files locally, you can download them from the AlphaPulldown Features Database, which contains precomputed protein features for major model organisms.

You can browse the full list and download individual features at https://alphapulldown.s3.embl.de/index.html or https://s3.embl.de/alphapulldown/index.html.

Installation

Note

For EMBL cluster users: You can access the directory with generated features files at /g/alphafold/input_features/

To access the Features Database, you need to install the MinIO Client (mc).

Steps:

  1. Download the mc binary.
  2. Make the binary executable.
  3. Move it to your PATH for system-wide access.

Example for AMD64 architecture:

curl -O https://dl.min.io/client/mc/release/linux-amd64/mc
chmod +x mc
sudo mv mc /usr/local/bin/

Verify installation:

To ensure mc is correctly installed, you can run:

mc --help

Configuration

Set up an alias for easy access to the AlphaPulldown Features Database hosted at EMBL:

mc alias set embl https://s3.embl.de "" "" --api S3v4

This alias allows you to interact with the Features Database as if it were a local directory.

Downloading Features

Once mc is installed and configured, you can start accessing the Features Database. The mc commands mimic standard bash commands.

List available organisms:

To view the list of available organisms with precomputed feature files, run:

mc ls embl/alphapulldown/input_features

Each organism directory has a features/ subdirectory holding compressed .pkl.xz feature files named by UniProt ID, each with a *_feature_metadata_<date>.json.xz sidecar recording the databases and software used. AlphaPulldown reads the compressed files directly — no need to decompress them.

Download specific protein features:

For example, to download the feature file for the protein with UniProt ID Q6BF25 from Escherichia coli, use:

mc cp embl/alphapulldown/input_features/Escherichia_coli/features/Q6BF25.pkl.xz Q6BF25.pkl.xz

Download all features for an organism:

To download all feature files for proteins from a specific organism, such as E. coli, copy the entire directory:

mc cp --recursive embl/alphapulldown/input_features/Escherichia_coli/ ./Escherichia_coli/

Alternatively, you can mirror the contents of the organism’s directory, ensuring all files are synced between the source and your local directory:

mc mirror embl/alphapulldown/input_features/Escherichia_coli/ Escherichia_coli/

This command mirrors the remote directory to your local system, keeping both locations in sync.

AlphaFold3 features

The database also hosts AlphaFold3 input features, kept separate from the AlphaFold2 set:

https://alphapulldown.s3.embl.de/input_features_af3/index.html

These are AlphaFold3 JSON inputs (*_af3_input.json.xz) with MSAs and templates embedded. The AlphaPulldown feature-generation metadata travels inside each file, in the description field of the first polymer entry, so there is no separate metadata sidecar. AlphaPulldown 2.6.2 and later read and write these compressed files directly.

mc ls embl/alphapulldown/input_features_af3
mc mirror embl/alphapulldown/input_features_af3/Bacillus_subtilis/features/ Bacillus_subtilis/
Organism Features Contributed by
Bacillus subtilis (UP000001570) 4,274 Christoph Elfmann, Stülke Lab, Georg-August-Universität Göttingen

Community contributions are welcome — please get in touch if you have a proteome to share.

Clone this wiki locally