Welcome to the homepage of FRTpred

    In this study, we proposed a stacking framework-based approach called FRTpred that predicts protein logarithmic folding rate constant ln(k_f) and folding type from the provided sequence. Firstly, 30 base-models (regression models for folding rate and classification models for folding type) were constructed by integrating 10 representative feature extraction methods and three commonly used machine-learning algorithms. Subsequently, the predicted values of 30 base-models were combined and inputted to random forest algorithm and constructed the final prediction model. Cross-validation analysis showed that FRTpred achieved mean absolute deviation of 1.491, 2.016, and 1.954 for non-two-state (N2S), two-state (2S), and combined (2S+N2S) ln(k_f) prediction. Moreover, it predicts folding type with an accuracy of 0.843. Performance comparisons based on independent test against the existing methods showed that FRTpred is more precise and promising in both ln(k_f) and folding type prediction.