Open Source Drug Discovery Stack

Refua at AgentCures

AgentCures builds in public at github.com/agentcures. Our mission is simple: open source to solve all human disease. Start with Refua, then layer in refua-mcp and refua-notebook for agent workflows and notebook-first research.

Open GitHub Install Refua

Products

Three key repositories in this stack:

refua

The core Refua package for running structure and affinity workflows directly from Python.

  • Move from target and ligand inputs to model outputs quickly
  • Base layer used across the Refua ecosystem

pip install refua

View repository

refua-mcp

An MCP bridge that lets assistant tools call Refua workflows in a structured way.

  • Designed for MCP-compatible clients
  • Great for orchestration and repeatable automation loops

pip install refua-mcp

View repository

refua-notebook

A Jupyter-first package for interactive, visual exploration of Refua objects.

  • Automatic widget rendering with %load_ext refua_notebook
  • ADMET, Molstar, SMILES, protein, small-molecule, and complex views

pip install refua-notebook

View repository

Quickstart

pip install refua
pip install refua-mcp
pip install refua-notebook
pip install "refua[cuda]"
from refua import Boltz2

model = Boltz2()
complex_spec = (
    model.fold_complex()
    .protein("A", "MKTAYIAKQRQISFVKSHFSRQLEERLGLIEVQ")
    .ligand("L", "CCO")
)

affinity = complex_spec.get_affinity()
print(affinity.ic50, affinity.binding_probability)

Build in Public

Use the repos, open issues, ship fixes, and help us push open source to solve all human disease.

Explore refua-mcp Explore refua-notebook