Boltz-2 biomolecular structure and binding affinity model weights
- Provider
- Boltz (Passaro, Corso, Wohlwend et al.)
- License
- MIT (source)
- Size
- 5 files
- Stored at
- Hugging Face (https://huggingface.co/boltz-community/boltz-2)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Drug Discovery
- Source
- https://www.biorxiv.org/content/10.1101/2025.06.14.659707v1
About Boltz-2 biomolecular structure and binding affinity model weights
Open weights for Boltz-2, the biomolecular foundation model from the Boltz team (MIT and Recursion) that predicts the 3D structure of protein, nucleic acid and small-molecule complexes and, for protein-ligand pairs, a binding affinity estimate. The mounted repo holds boltz2_conf.ckpt (the structure and confidence model, 2.3 GB), boltz2_aff.ckpt (the affinity head, 2.1 GB) and mols.tar (the chemical component dictionary molecules the model needs at inference, 1.9 GB). Point the boltz package's cache at the mount to skip the download, then run predictions on inputs built from the PDB, ChEMBL or BindingDB archives. Code and weights are MIT licensed.
How to load Boltz-2 biomolecular structure and binding affinity model weights in Python
import os, json
base = "/mnt/archive" # this model repo, mounted read-only
files = sorted(os.listdir(base)); print(files)
print("boltz2_aff.ckpt", os.path.getsize(os.path.join(base, "boltz2_aff.ckpt")) // 2**20, "MB")Written for a Scigantic notebook, where /mnt/archive is the mounted dataset.
Data structure
Files
Example file: boltz2_aff.ckpt
Analyze this dataset without downloading it
On Scigantic, Boltz-2 biomolecular structure and binding affinity model weights opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.
Related Drug Discovery datasets
- Consensus QSAR models estimating acute aquatic toxicity for three trophic levels organisms: Algae, Daphnia and Fish
- An assessment of the (anti)androgenic properties of hexachloronaphthalene (HxCN) using a model of immature male rats (Hershberger Bioassay)
- Data from: Covalent disruptor of YAP-TEAD association suppresses defective Hippo signaling
- Normalized spacial scores of COCONUT compounds
- Automated benchmarking of combined protein structure and ligand conformation prediction
- Transformations in PubChem - Full Dataset
- OpenADMET CYP Inhibition Blind Challenge Dataset
- Systematic Data Analysis and Diagnostic Machine Learning Reveal Differences between Compounds with Single- and Multitarget Activity