Date of Award
3-18-2009
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Computer Science
First Advisor
Peiyi Tang
Abstract
Protein structure prediction is one of the most important problems in molecular biology. Secondary structure prediction is the initial starting point in determining three-dimensional structure of a protein. This work presents a novel approach to predict protein secondary structure using Markov chain Monte-Carlo simulation. The transition probabilities are first data-mined from the Dictionary of Protein Secondary Structure which contains over 46,000 known proteins. Monte-Carlo simulation uses these transition probabilities to generate thousands of secondary structure sequences for a given sequence of amino acids. Among these simulated secondary structure sequences, the one with the highest probability is chosen as the prediction. Several clustering algorithms have been applied to identify a sequence of the highest probability. The two most successful of which are presented in this thesis. First, Exact match frequency clustering has successfully been applied to relatively short proteins (up to 100 amino acids). Second, Sov-based clustering, clusters sequences based on the Segment Overlap measure and presents potential to reach the highest limit of prediction accuracy possible. The experimental runs show that the proposed approach achieves slightly higher per-residue accuracy than the neural network algorithm NNPredict. This thesis contains the background information, theory, algorithm design, implementation and experimental results of the proposed approach.
Recommended Citation
Im, Inessa G., "Predicting Protein Secondary Structure Using Markov Chain Monte-Carlo Simulation" (2009). Theses and Dissertations. 233.
https://research.ualr.edu/etd/233
