An Ensemble Classifier with Random Projection for Predicting Protein-Protein Interactions Using Sequence and Evolutionary Information

摘要: Identifying protein-protein interactions (PPIs) is crucial to comprehend various biological processes in cells. Although high-throughput techniques generate many PPI data for species, they are only a petty minority of the entire network. Furthermore, these approaches costly and time-consuming have high error rate. Therefore, it necessary design computational methods efficiently detecting PPIs. In this study, random projection ensemble classifier (RPEC) was explored identify novel PPIs using evolutionary information contained protein amino acid sequences. The obtained from position-specific scoring matrix (PSSM) generated PSI-BLAST. A feature fusion scheme then developed by combining discrete cosine transform (DCT), fast Fourier (FFT), singular value decomposition (SVD). Finally, via classifier, performance presented approach evaluated on Yeast, Human, H. pylori datasets 5-fold cross-validation. Our achieved prediction accuracies 95.64%, 96.59%, 87.62%, respectively, effectively outperforming other existing methods. Generally speaking, our quite promising supplies practical effective method predicting