Makeshift: a lightweight software for accessing and analyzing NMR data and protein dynamics

Nuclear magnetic resonance (NMR) spectroscopy generates detailed molecular data and insights into chemical environments at the atomic level, serving as a foundation for predictive models in biochemistry. Extensive NMR datasets are stored in the public Biological Magnetic Resonance Data Bank, but accessing and interpreting this information at scale in computational workflows has proven challenging.

The researchers introduce makeshift, an open-source Python package designed to facilitate the retrieval, curation, and analysis of NMR data. Users can easily access BMRB entries and execute key analyses, including chemical shift re-referencing, secondary structure propensity prediction, and the interpretation of relaxation datasets for understanding biomolecular dynamics.

The researchers also reimplemented several widely-used NMR calculations that were either not open-source or unavailable in Python, validating their implementations against the original methods. By unifying data access, processing, and analysis within a single Python interface, makeshift significantly reduces the barriers to reproducible, scalable analysis and machine learning applications utilizing biomolecular NMR data.