Fine adjustments of thermo-vibrations between residues surrounding the active center in protein using dual artificial intelligence approaches and computer simulations

Adjustments of thermo-vibrations in proteins should be anticipated to provide novel helpful understanding of its properties. R111 and Q112 in the strand structure between the two domains of papain were repeatedly replaced to increase the thermo-vibration between residues surrounding its active center. This occurs due to collaborative activities composed of selections by deep neural network, verification by molecular dynamics simulations, and trainings of deep neural network by the verification results. The three replacements (R111Y-Q112K, R111C-Q112N, and R111W-Q112E) were found to increase the thermo-vibration and were expected to attain higher-temperature characteristics while at a normal temperature. The decision tree highlighted aromatic carbon and oxygen atoms in the 111th amino acid and carbon and nitrogen atoms in the 112th amino acid as influential factors. These atoms should be kept in mind to effectively carry out the collaborative activities.