OsmoFold: A high-throughput tool for predicting the impact of osmolytes on protein structure

Small molecules can affect a protein's fold and stability. Physical chemistry conceptualizes this effect using Tanford's transfer model, which describes the change in free energy of a protein upon transfer from water to a solution containing some concentration of a small molecule. Using additive transfer free energies of individual amino acids, one can apply the transfer model to quantitatively predict how small molecules affect protein folding equilibria. However, no open-source framework currently exists for applying this method at scale. Here, we present OsmoFold, an open-source suite of tools designed to act as a user-friendly, high-throughput option for predicting the impact of osmolytes on protein folding. OsmoFold is available as a Google Colab notebook, requiring no programming experience and only a protein structure file as input, and as a Python package for the analysis of large data sets and seamless integration into bioinformatic pipelines. To showcase OsmoFold's versatility, we demonstrate its ability to recapitulate results from published experimental data. In addition, we show OsmoFold's utility in predicting ideal excipients for biologic pharmaceuticals using an in vitro study of human blood clotting factor VIII. Our results reaffirm the utility of this framework for predicting the effect of osmolytes on protein stability. At the same time, the accompanying tool makes this framework more accessible and versatile than before. In doing so, OsmoFold has the potential to drive advancements in protein stability research, improving biologic formulation strategies and therapeutic development.