About

Tuberculosis (TB) is one of the most epidemic diseases all around the world. its serious consequences have caused high concern in the World Health Organization. Researches show that identification of secretory protein antigen of mycobacterium tuberculosis (M. tuberculosis) is an available approach of detecting infected individuals. To identify the secretory protein of mycobacterium tuberculosis in bioinformatics way.

In this work, we emphasised an imbalance data strategy and proposed a SVM-based identification model. In this model, The improved PseAAC algorithm was used to extract feature and the imbalanced data strategy was applied to get train set and test set. Subsequently, we adopted SVM classifier without feature selection to construct and optimize the final model, which achieved the 86% of AA( Average Accuracy) and 0.862 of AUC in independent test set. Based on this model, online server SecProMTB was constructed