Prediction of drug classes based on gene expression data

Abstract

Nowadays, the financial investments in pharmaceutical research and development are an enormous increase. Drug safety is very important to health and drug development. Finding new uses for the approved drug has become important for the pharma industry. Drug classification accuracy helps identify useful information for studying drugs, also helps in accurate diagnosis of drugs. Gene expression data makes a possible study of biological problems and machine learning methods are playing an essential role in the analysis process. Meanwhile, many machine learning methods have been applied to classification, clustering, dynamic modeling areas of gene expression analysis.

This thesis work is using R programming language and SVM machine learning method to predict the ATC class of drugs based on the gene expression data to see how well the gene expression patterns correlate after treatment within the therapeutic/pharmacological subgroup. A dimensionality reduction method will use to reduces the dimensions of the dataset that improves the classification performance. The classifiers built using SVM machine learning technique in this thesis study had limited with detecting drug groups based on the ATC system.