repository·main·Indexed 43 days ago
https://github.com/openfreeenergy/openfeA Python package for planning and executing alchemical free energy calculations to support molecular simulation in drug discovery. It provides a core library for defining LigandNetworks and AlchemicalNetworks, along with a CLI for network planning (e.g., plan-rbfe-network), execution via quickrun, and result gathering.
openfe is a Python package designed for planning and executing alchemical free energy calculations. It is the flagship project of the Open Free Energy consortium, which provides open-source tools for molecular simulation in drug discovery.
Key resources:
OpenFE is a free and open-source Python framework designed for performing alchemical free energy calculations. It provides a unified toolkit to plan, execute, and analyze these calculations using a variety of different methods.
Key resources for developers include:
OpenFE documentation covers simulation and analysis workflows:
cinnabar Python package to analyze results, such as generating Maximum Likelihood Estimation (MLE) estimates of absolute free energies and plotting free energy networks.OpenFE supports absolute free energy protocols:
OpenFE provides several ways to perform Relative Binding Free Energy (RBFE) calculations:
An AlchemicalNetwork is the central object used to plan a simulation campaign. It contains all the necessary information to execute a series of free energy calculations, including the chemical systems involved and the alchemical transformations between them.
An AlchemicalNetwork is modeled as a graph:
ChemicalSystem objects, which describe specific molecules.Transformation objects, which define how to perform the simulation between two nodes.In a valid network, every node must be associated with at least one transformation (edge); there are no disconnected or solitary nodes. A Transformation includes the information for the two ChemicalSystem nodes it connects, a Protocol, and (for alchemical transformations) atom mapping information, often managed via a LigandNetwork.
A LigandNetwork is a graph-based representation used to organize free energy calculations for a set of molecules.
SmallMoleculeComponent objects (e.g., drug candidates).LigandAtomMapping objects, which define how atoms correspond between molecules.LigandNetwork objects are immutable once created. You can associate a score with each LigandAtomMapping, which network generators use to optimize the network topology (e.g., creating a more efficient path for calculations).
A Protocol is a computational method used to estimate the free energy difference between two chemical systems. OpenFE provides several built-in protocols, including:
RelativeHybridTopologyProtocolAbsoluteBindingProtocolSepTopProtocolAbsoluteSolvationProtocolPlainMDProtocolBecause all Protocol objects share a common interface for creation and execution, you can easily swap methods or benchmark different protocols for your project.
The SepTopProtocol is used to calculate the difference in binding free energy between two chemically diverse ligands. Unlike hybrid topology protocols, SepTopProtocol treats the two ligand topologies as completely separate, meaning no common core is required and atom mapping is unnecessary.
It works by performing two absolute binding free energy calculations simultaneously in opposite directions: alchemically inserting one ligand into the binding site while removing the other. The relative binding free energy is obtained via a thermodynamic cycle involving both the complex (binding site) and solvent phases.
LigandNetwork represents a network of mutations between ligands. It is a core abstraction used in openfe to define the pathways of chemical transformations (mutations) that connect different small molecules, typically used to plan free energy perturbation (FEP) workflows.