Development of Computer-Aided Rational Protein Engineering Toolkit (CARPET)

Protein engineering is varying the structure of a protein to develop valuable proteins. Protein side-chain mutation is fundamental to protein engineering processes, but the astronomical number of all possible protein sequences makes protein engineering difficult. Thus, computer-aided protein engineering is required to explore all possible protein sequence space and to guide the effective protein design. A number of different approaches have been suggested and applied to prediction of protein properties, such as enzyme reactivity, binding affinity, substrate/product selectivity, thermal/pH stability, and solubility, through sequence-based machine learning algorithms, knowledge-based potential functions, all-atom molecular mechanics based calculations, and all-atom quantum mechanics based calculations. Here, we suggest a new Computer-Aided Rational Protein Engineering Toolkit (CARPET) which predicts protein properties via sequence-based algorithms and structure-based algorithms, and helps rational enzyme modification. In CARPET, it starts from phylogenetic analysis via structural superimposition for inter-species comparison. The good score mutations are selected from the suggested mutation pool based on prediction models of thermal stability and solubility. Free Energy Perturbation (FEP) and Thermodynamic Integration (TI) are used to calculate more accurate thermal/pH stability. Finally, fragment molecular orbital (FMO) method is used to calculate calculate residue-residue interactions in enzyme to predict protein properties. Through collaborative manners in our new platform, we can analyze the experimental mutation results, design putative mutation pool, predict the properties of suggested mutation, and finally pick out the best-score mutations.