Computational studies for Medical Profiling

Abstract

Hepatitis, an inflammatory disease of liver, is one of the commonest infectious disease of third world nations. Current treatment strategies are reported to cause adverse side effects along with drug resistance. Post Human Genome era has provided vast inputs of information pertaining to the infective agents. By amalgamation of various techniques of advanced computational, mathematical and statistical analysis, the pattern mining based prediction studies are being undertaken for understanding the biological or medical data. The need of the hour is to study patterns which can lead to improved understanding with respect to genotypic variation of the virus, inhibitory antiviral drug activity and clinical outcome. newlineWork presented in my thesis addresses three major aspects of the aforementioned problem: (1) phylogenetic study (2) prediction of compound activity based on the available physicochemical data and (3) predicting medical outcomes. The thesis work was carried out in five parts. Chapter 1 dealt with analysis of the available Hepatitis medical dataset for finding improved event prediction model as an aid to supervised clinical decision making. Chapter 2 was focused on the phylogenetic analysis of hepatitis E virus where comparative analysis of the evolutionary aspect of the organism was performed using seven alignment free methods and complete genotype based study was performed using the Return Time distribution (RTD) based alignment free method. As a promising future therapy, Chapter 3, 4 and 5 were dedicated to antimicrobial peptides, which were studied for attributes which can help in better classification and predictor of antiviral activity in context of anti-hepatitis peptides. newline Chapter 3 focused on the study of the use of algorithms and support vector machines as classifier to improve modeling on the available hepatitis antimicrobial dataset. We also aimed to expand the study on antiviral peptides for wider application and compile it under new database as described in Chapter 4 . Chapter 5 included