PROtein reSidue Annotation Toolkit
* ProSAT_Learn is the program to learn "n" one-versus-rest classification models, where n are the possible number of annotations. It can also learn a support vector regression model to predict a floating point value.
* ProSAT_Predict is the program to predict the annotation for each residue from the built models, and output a L x n dimension profile, outputting the score from each of the n one-versus-rest SVM models for each amino acid residue and L is the length of the protein sequence. In case of a regression model, n =1
* ProSAT_Eval is the program to evaluate the various parameters like kernel, window lengths, and regularization parameters. ProSAT_Eval automatically performs cross validation and reports the standard classification and regression performance. Please use the perl script cv_script.pl to automatically perform model selection and training the best classification or regression model for the problem.