BioForge: A Fast and Safe Macromolecular Structure Preparation Library in Rust

Abstract Summary BioForge is the first macromolecular structure preparation library for Rust, complementing sequence-analysis libraries such as Rust-Bio with cleaning, geometry-based missing-atom repair, protonation, and solvation as a force-field-agnostic preprocessing stage. Across 22 structures (1439–3 819 839 atoms) on a single CPU core, BioForge scales linearly in wall time (fitted N0.89) versus super-linear N1.17 for PDBFixer and N1.25 for PDB2PQR, completing the 3.8M-atom 9MJN in 5.1 s at 1.1 GB peak (PDBFixer: error, 273 GB); PDB2PQR is limited to fewer than 100 000 atoms. BioForge also compiles to WebAssembly for browser deployment and supports direct embedding into simulation and parameterization pipelines via its application programming interface. Availability and Implementation An MIT-licensed Rust library crate (bio-forge) and a WebAssembly package (bio-forge-wasm) extend the Rust bioinformatics ecosystem with structure preparation, both published on https://crates.io and NPM respectively; the same WebAssembly binary powers a client-side web application at https://bio-forge.app that runs entirely in the browser without a server or data upload. Source code and precompiled command-line binaries are freely available at https://github.com/TKanX/bio-forge; an archival snapshot of the version used for all reported experiments is deposited at Zenodo (https://doi.org/10.5281/zenodo.21627953). Supplementary Information Supplementary data are available at Bioinformatics online.