Protein structure prediction (RMSD ≤ 5 Å) using machine learning models

Physical and chemical properties of protein help to determine the quality of protein structure. Here we explore the machine learning models using six physical and chemical properties, namely total empirical energy, secondary structure penalty, total surface area, pair number, residue length and Euclidean distance to predict the RMSD of a protein structure in the absence of its true native state. The Real Coded Genetic Algorithm is used to determine feature importance, and k-fold cross-validation is used to measure the robustness of the best predictive model. The experiments show that the random forest model outperforms the other machine learning approaches in RMSD prediction. The performance result shows that in the prediction of RMSD, the Root Mean Square Error (RMSE) is 0.48, correlation is 0.90, R² is 0.82 and accuracy is 97.02% (with ±2 error) on the testing data. The data set used in the study is available at http://bit.ly/PSP-ML.