Dissertation/Thesis Abstract

Protein 3-D Structure Representation for Alignment-Free Comparison and Structural Motif Discovery of Proteins, and Hierarchical Protein Classification
by Singh, Sumi, Ph.D., University of Louisiana at Lafayette, 2015, 140; 10003582
Abstract (Summary)

Proteins are made of amino acid sequences that fold into complex 3-D structures. These 3-D structures make proteins functional therefore, structure comparison provides a more sensitive way of homology database search. Two of the greatest challenges towards comparing protein structures is attributable to the complexity of the protein macro-molecule in addition to its size. To add to the complexity is the fact that, while obtaining the physical coordinates for each atom, the reference frame selected is usually based on the most stable atom/amino acid, resulting in an arbitrary reference frame based coordinate system.

The most popular 3-D structure comparison algorithms are computationally expensive, distance-based methods that require translation and rotation to achieve perfect alignment, before calculating the root mean square distance between the two macro molecules. Furthermore, adding to the limitations of these algorithms is their inability to identify local structural similarity between proteins having very different global structures.

In this work, I propose an algorithm for 3-D protein structure representation based on Triangular Spatial Relationships (TSR) between the structural constituents (amino acids) of the proteins. These structural representations act as structural fingerprints and are invariant to rotation and translation of the protein macro-molecule and, therefore, do not depend upon the choice of the reference frame. The structural fingerprints can be used to compare two structures locally and globally. In the dissertation, a novel, global structural comparison method using TSR 3-D, is introduced. TSR 3-D has been experimentally validated to show comparable results with respect to some other standard methods, but with far less computation. A local structural comparison method to obtain insight into structural motifs is also proposed. These structural motifs are responsible for functional similarity between different proteins. TSR 3-D-based functional/structural motif discovery is shown to give better results when compared to the standard, sequence-based methods. TSR 3-D-based structural features are shown to generate better hierarchical classification results on protein SCOP classification, when compared to the commonly used flat classification. Finally a tool based on a NoSQL database using TSR 3-D has been designed for fast protein structure comparison.

Indexing (document details)
Advisor: Raghavan, Vijay V.
Commitee: Chu, Chee-Hung Henry, Miao, Jin, Xu, Wu
School: University of Louisiana at Lafayette
Department: Computer Science
School Location: United States -- Louisiana
Source: DAI-B 77/06(E), Dissertation Abstracts International
Subjects: Biochemistry, Bioinformatics, Computer science
Keywords: Hierarchical classification, Nosql mongodb, Protein structure comparison, Proteomics, Structural motif, Triangular spatial relationships
Publication Number: 10003582
ISBN: 9781339426983
Copyright © 2019 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy