Optimal Classification on Selected Genes With Gene Ontology and Protein Analysis

Abstract

In the rapidly evolving field of genomic research, their arise so much of highly potential data due to the advances in modern technological biomedical instrumentation. So the collection and processing of this information is highly sophisticated and challenging task. One can use multi-layer novelistic approaches in the field of data mining and soft computing to extract that disease gene information, also by further processing the same information using Machine Learning (ML) algorithms and pattern recognition methods one can retrieve more information for analysis. newlineIn the emerging field of ML, feature selection and classification makes it possible to classify the distinct classes of disease samples. In this method gene selection is performed on the disease sample high dimensional gene data set to extract the most highly expressed genes using statistical filter based approaches and by using thus obtained significant gene data expression, one can train the coined ML approaches to learn and classify the disease samples from normal samples and to measure the accuracy of classifier algorithms. It uses four continuous gene selection methods such as Pearson Correlation Coefficient (PCC), Signal to Noise Ratio (SNR), Feature Assessment by Sliding Threshold (FAST) , and Feature Assessment by Information Retrieval (FAIR) and by using Support Vector Machine (SVM), K- Nearest Neighbour (KNN) and Naive Bayesian (NB) as classifiers one can measure the performance of various gene selection methods acting on various classifiers. On classification, SNR with NB classifier shows high accuracy for most of disease datasets. newlineIn the growing field of Gene Ontology (GO) analysis, genes are arranged based on numerous Gene Interaction Network links, with set of gene terms associated for each gene. By using this information, one can associate two or more genes and find the similarity of their GO terms using certain hierarchical clustering. In this gene selection based GO analysis approach, the gene features are selected using clustering and