GIST-gpu

GIST-gpu is a GPU-accelerated implementation of the grid inhomogeneous solvation theory (GIST) algorithm, which calculates the thermodynamic properties of water molecules surrounding a protein. The original GIST algorithm is computationally demanding, especially for large systems, with calculations taking days or weeks. GIST-gpu addresses this issue by leveraging the parallel processing capabilities of GPUs, enabling efficient estimation of solvation-free energy for large biomolecular interfaces.

The authors demonstrate that GIST can be a reliable tool for evaluating protein surface hydrophobicity. They apply GIST-gpu to a set of nine different proteases, calculating localized solvation-free energies on the surface of the binding interfaces as a measure of their hydrophobicity. The results show a strong agreement between the binding interfaces' calculated hydrophobicity and their substrates (peptides) that bind into the binding cleft, suggesting that GIST-GPU provides a reliable description of the hydrophobicity characteristics of biological interfaces.