|
||||||
|
||||||
| Undergraduate Honours Projects | ||||||
|
Carleton University - School of Computer Science Undergraduate Honours Project Fall 2012 Bioinformatics: Algorithms and Methods in Protein Structure Prediction Wiktor Kula 
ABSTRACT Protein-structure prediction problem is an open problem in Bioinformatics and the increased attempt to resolve this problem has led to the development of various approaches. This report examines the algorithms and methods in protein structure prediction. The idea that accurate prediction of protein structure facilitates the elucidation of their biochemical functions motivates this work. Various methods for predicting the three-dimensional (3D) structure of proteins have been developed and can be categorized into comparative modeling methods, fold recognition and methods based on the first principles with and without database information. Generally, protein-structure prediction techniques envisage protein conformations using knowledge on the primary amino acid sequences. The homology modeling method uses homology techniques to resolve the protein-structure prediction problem by making comparisons with known protein structures. The Dot Matrix algorithm seeks to solve the structure prediction problem by comparing two similar sequences to estimate homology, while Dynamic Programming (DP) provides an optimal alignment for pair-wise characters. Similarly, Linear Programming (LP) algorithms are robust tools for solving the linear feasibility problem and the SVM (Support Vector Machine) approach involves solving the burdensome quadratic programming problem. A critical analysis of the existing approaches suggests that the boundaries between them are becoming blurred with increased sophistication of the various algorithms and methods. |
||||||
|
