The article presents the vermouth python library and the martinize2 program, which together provide a unified framework for developing programs to prepare, run, and analyze coarse-grained molecular dynamics (MD) simulations using the Martini force field.
The vermouth library defines an API with data structures and independent processes to support various workflows commonly encountered in Martini programs. It separates the stages of topology generation into reading input, identifying and repairing atoms, mapping to coarse-grained resolution, generating bonded interactions, and post-processing. This modular design allows for better code quality, robustness, and extensibility.
Martinize2 is built on top of the vermouth library and serves as the successor to the previous martinize script. It can automatically handle protonation states, post-translational modifications, and the conversion of non-protein molecules such as ligands. Martinize2 also offers more options to fine-tune the elastic network used to maintain the tertiary structure of proteins.
The authors demonstrate the capabilities of martinize2 by processing the entire I-TASSER protein template database and a subset of the AlphaFold Protein Structure Database. The results show that martinize2 is more robust than the previous martinize script, with the ability to detect and handle problematic input structures, while still being efficient enough for high-throughput applications.
다른 언어로
소스 콘텐츠 기반
arxiv.org
더 깊은 질문