Date of Submission

5-2019

Document Type

Thesis

Degree Name

Master of Science in Computer Science

Department

Electrical & Computer Engineering and Computer Science

Advisor

Said Mikki

Committee Member

David Eggert

Committee Member

Saion Sinha

MeSH

Biosensing Techniques, Machine Learning, DNA Probes

LCSH

Biosensors, Machine learning

Abstract

In molecular biology, the term “DNA hybridization” generally refers to the process of forming a double stranded nucleic acid from joining two complementary strands of DNA. The degree of genetic similarity of the DNA resulting from hybridization can be detected ei ther by using the chemical characteristics of DNA samples or by utilizing reliable biosensors which transform the chemical characteristics into a source of electrical measurements. In past research about such sensors, known as DNA Hybridization Detection Systems, the thermal and electrical characteristics of carbon nanotubes are utilized to detect whether hybridization takes place or not. However, human interpretation of the measured data can lead to uncer tainty regarding, which compromises one crucial characteristic of biosensors—reliability. Research aimed at greater understanding of this sensor is still very much underway. This study is intended to make a significant contribution to the growing field of biotechnology by means of analyses of machine learning methods from a classification perspective. The ultimate goal of this thesis is to see if any of the existing classification algorithms, such as k-nearest neighbors and decision trees, are capable of predicting the state of hybridization based on simple electrical measurements. In this way, the machine learning algorithm uses real-life data to provide a systematic tool that can be utilized in various fields, such as bioinformatics, biomedicine, and forensic science.

Available for download on Friday, October 09, 2020

Share

COinS