Prediction of Protein pKa with Representation Learning

The behavior of proteins is closely related to the protonation states of the residues. Therefore, prediction and measurement of pKa are essential to understand the basic functions of proteins. In this work, we develop a new empirical scheme for protein pKa prediction that is based on deep representation learning. It combines machine learning with atomic environment vector (AEV) and learned quantum mechanical representation from ANI-2x neural network potential (J. Chem. Theory Comput. 2020, 16, 4192). The scheme requires only the coordinate information of a protein as the input and separately estimates the pKa for all five titratable amino acid types. The accuracy of the approach was analyzed with both cross-validation and an external test set of proteins. Obtained results were compared with the widely used empirical approach PROPKA. The new empirical model provides accuracy with MAEs below 0.5 for all amino acid types. It surpasses the accuracy of PROPKA and performs significantly better than the null model. Our model is also sensitive to the local conformational changes and molecular interactions.

Protein pKa prediction by tree-based machine learning

We present four tree-based machine learning models for protein pKa prediction. The four models, Random Forest, Extra Trees, eXtreme Gradient Boosting (XGBoost) and Light Gradient Boosting Machine (LightGBM), were trained on three experimental PDB and pKa datasets, two of which included a notable portion of internal residues. We observed similar performance among the four machine learning algorithms. The best model trained on the largest dataset performs 37% better than the widely used empirical pKa prediction tool PROPKA. The overall RMSE for this model is 0.69, with surface and buried RMSE values being 0.56 and 0.78, respectively, considering six residue types (Asp, Glu, His, Lys, Cys and Tyr), and 0.63 when considering Asp, Glu, His and Lys only. We provide pKa predictions for proteins in human proteome from the AlphaFold Protein Structure Database and observed that 1% of Asp/Glu/Lys residues have highly shifted pKa values close to the physiological pH.

Protein pKa prediction with machine learning

Protein pKa prediction is essential for the investigation of pH-associated relationship between protein structure and function. In this work, we introduce a deep learning based protein pKa predictor DeepKa, which is trained and validated with the pKa values derived from continuous constant pH molecular dynamics (CpHMD) simulations of 279 soluble proteins. Here the CpHMD implemented in the Amber molecular dynamics package has been employed (Huang, Harris, and Shen J. Chem. Inf. Model. 2018, 58, 1372-1383). Notably, to avoid discontinuities at the boundary, grid charges are proposed to represent protein electrostatics. We show that the prediction accuracy by DeepKa is close to that by CpHMD benchmarking simulations, validating DeepKa as an efficient protein pKa predictor. In addition, the training and validation sets created in this study can be applied to the development of machine learning based protein pKa predictors in future. Finally, the grid charge representation is general and applicable to other topics, such as the protein-ligand binding affinity prediction.