In this post, we highlight research published in the Journal of Chemical Information and Modeling that demonstrates how OpenBioSim’s software stack enables assembly of workflows for systematic comparison of different computational approaches for predicting protein-ligand binding affinities.
Why benchmarking matters
Alchemical free energy methods are gaining significant traction in computer-aided drug discovery. An expanding array of methodologies is now available for the setup, execution, and analysis of relative binding free energy (RBFE) calculations. Different research groups and organizations have developed various algorithms and protocols, each with their own strengths and optimal use cases. However, comparing these approaches has historically been challenging due to incompatible software, outdated file formats, and inconsistent methodologies. This lack of standardization makes it difficult for practitioners to determine which methods work best for their specific drug discovery projects.
Leveraging BioSimSpace for interoperable benchmarking
To address this challenge, a team of researchers from the University of Edinburgh and the company Evotec leveraged OpenBioSim’s BioSimSpace framework to build modular and interoperable RBFE workflows. BioSimSpace’s interoperability features were essential for this study, as they enabled the team to combine different setup tools, simulation engines, and analysis methods within unified workflows. This approach allowed for fair, systematic comparisons that would have been impractical or impossible using traditional software-specific pipelines.
The study assessed the performance of various community-developed setup, simulation, and analysis tools on a benchmark set of six protein-ligand congeneric series. By testing multiple combinations of tools across consistent test systems, the researchers were able to recommend best practices for the reliable application of RBFE methods in drug discovery.
Key contributions of OpenBioSim software
BioSimSpace served as the backbone for creating modular workflows where individual components could be swapped and tested systematically. The framework’s ability to interface with multiple molecular dynamics engines and analysis tools meant that researchers could evaluate different methodological choices without needing to rebuild entire workflows from scratch. This modularity is precisely what enables comprehensive benchmarking studies that can guide the broader computational chemistry community toward more reliable predictions.
The interoperable nature of BioSimSpace addresses a fundamental barrier in computational drug discovery: the difficulty of sharing and comparing methodologies across different software ecosystems. By providing a common interface layer, BioSimSpace empowers researchers to focus on scientific questions rather than wrestling with incompatible file formats and software dependencies.
Advancing best practices in drug discovery
This benchmarking study provides valuable recommendations on best practices that can help drug discovery teams achieve more reliable RBFE calculations. By systematically evaluating multiple approaches, the research identifies which combinations of tools and protocols deliver the most accurate and robust results. Such guidance is invaluable for both academic researchers and pharmaceutical companies seeking to incorporate alchemical free energy methods into their drug discovery pipelines.
The work exemplifies how open-source, interoperable software infrastructure can accelerate scientific progress by enabling rigorous comparisons that benefit the entire community. As alchemical free energy methods continue to mature and find wider application in drug discovery, systematic benchmarking studies like this one will be important for establishing standards and best practices.
References
Herz, A. M.; Bissaro, M.; Esposito, C.; Michel, J. “Modular and Interoperable Workflows for Benchmarking Alchemical Binding Free Energy Calculation Methodologies” Journal of Chemical Information and Modeling, 2025, 65 (19), 10658-10672 doi:10.1021/acs.jcim.5c01493
BioSimSpace paper: Hedges, L. O.; Mey, A. S. J. S.; Laughton, C. A.; Gervasio, F. L.; Mulholland, A. J.; Woods, C. J.; Michel, J. “BioSimSpace: An interoperable Python framework for biomolecular simulation” Journal of Open Source Software, 4(43), 1831, 2019 doi:10.21105/joss.01831
Written by Claude and edited by Julien Michel
