Open AccessClassification of Hot Spots using XGBoost and LightGBM Algorithms
Author(s) -
M. Vijayakumar,
Joby George
Publication year - 2020
Publication title -
international journal of engineering and advanced technology
Language(s) - English
Resource type - Journals
ISSN - 2249-8958
DOI - 10.35940/ijeat.e9459.069520
Subject(s) - support vector machine , boosting (machine learning) , computer science , classifier (uml) , matthews correlation coefficient , artificial intelligence , pattern recognition (psychology) , data mining , alanine scanning , machine learning , algorithm , spots , gene , biology , mutation , botany , biochemistry , mutagenesis
Protein-Protein Interactions referred as PPIs perform significant role in biological functions like cell metabolism, immune response, signal transduction etc. Hot spots are small fractions of residues in interfaces and provide substantial binding energy in PPIs. Therefore, identification of hot spots is important to discover and analyze molecular medicines and diseases. The current strategy, alanine scanning isn't pertinent to enormous scope applications since the technique is very costly and tedious. The existing computational methods are poor in classification performance as well as accuracy in prediction. They are concerned with the topological structure and gene expression of hub proteins. The proposed system focuses on hot spots of hub proteins by eliminating redundant as well as highly correlated features using Pearson Correlation Coefficient and Support Vector Machine based feature elimination. Extreme Gradient boosting and LightGBM algorithms are used to ensemble a set of weak classifiers to form a strong classifier. The proposed system shows better accuracy than the existing computational methods. The model can also be used to predict accurate molecular inhibitors for specific PPIs