Predicting Primary Sequence-Based Protein-Protein Interactions Using a Mercer Series Representation of Nonlinear Support Vector Machine

The prediction of protein-protein interactions (PPIs) is essential to understand the cellular processes from a medical perspective. Among various machine learning techniques, kernel-based Support Vector Machine (SVM) has been commonly employed discriminate between interacting and non-interacting protein pairs. main drawback employing SVM datasets with many features, such as primary sequence-based dataset, significant increase in computational time training stage. This mainly due presence kernel solving quadratic optimisation problem (QOP) involved nonlinear SVM. In order fix this issue, we propose novel efficient algorithm by approximating using low-rank truncated Mercer series well desired. As result, QOP for approximated will be very tractable sense that there reduction validating stages. We illustrate novelty proposed method predicting PPIs “S. Cerevisiae” where features extracted multiscale local descriptor (MLD), then compare predictive performance approximation existing methods. Finally, new results almost accuracy