caspredict

Source code related to Cas protein prediction. For the convenience of users, the SVM-based model built with ODPC was implemented into an online web service, called CASPredict, which is freely available at http://i.uestc.edu.cn/caspredict/cgi-bin/CASPredict.pl. All the source code related to model building is here. In the source code, we borrowed some of the resources in libsvm(Copyright (c) 2000-2021 Chih-Chung Chang and Chih-Jen Lin). License:Caspredict is developed and released under the MIT license. Folder Introduction: In "Model building", the user should first execute feature-extraction.py in the "feature-extraction" folder based on the input data. Secondly, according to the results of feature extraction, the feature selection strategies of F-score(F-score-feature-selection.py) and random forest(Random-forest-feature-selection.py) were respectively used for 10-fold cross validation. At the same time, DPC-without-feature-selection.py did not perform feature selection in the 10-fold cross-validation. Thirdly, comparing the results obtained by the ten-fold cross-validation, the accuracy of feature selection based on F-score was higher, so the feature selection was based on F-score in the leave-one-out cross-validation(leave-one-out-feature-selection.py). lastly, the Final-model-building.py was used for the construction of the final model.