FUNCTIONAL PROFILING AND CLASSIFICATION OF GENETIC VARIANTS

Abstract

The study of genetic variants in human DNA is sufficiently important in assisting researchers to discover the genetic contributions to many human diseases such as cancer, heart diseases, Alzheimer disease, and diabetes. A genetic variant, or variant, is a change in the DNA sequence that may have some negative and harmful outcomes. The single nucleotide polymorphism, SNP, is the most popular type of variant and caused by a change in one nucleotide base in mostly the protein-coding region of the DNA sequence. The main goal of this thesis is to study, analyze and classify human genetic variants based on their pathogenicity. In this task, the functional profiles of genetic variants are constructed and used in the methodology to determine whether a genetic variant is pathogenic, causing disease, or benign using biological process functional profiling from the Gene Ontology. The presented method in this thesis was trained on a set of 2,000 variants and tested with ten-fold cross-validation. The evaluation and experiment results show that this method is effective and suggest that predicting the pathogenicity from the Gene Ontology is significant. The scoring accuracy of our method is as high as 90.1 % by using 10-folds cross validation.