STUDIES ON CORRELATED MUTATIONS ALGORITHMS OF PROTEINS PROVIDING STRUCTURAL, SPATIAL, AND ALLOSTERY INFORMATION FROM MULTIPLE SEQUENCE ALIGNMENTS

Proteins provide innumerable cellular functions and benefits for all kingdoms in the domains of life. Advancements in the high throughput collection and analysis of proteins have led to ever-deeper understanding of biological pathways, evolution, and coding biases. Most protein functional and/or structural analysis that is carried out in an in vitro manner is not amenable to high throughput technologies. With the incredible growth of sequences to study, we have capabilities to further refine algorithms that work in silico, using the work done in vitro as a benchmark. There has been a renaissance of the study of proteins using new approaches that are largely possible because of the amount of data now available for analysis. The research in this dissertation investigates some of the new techniques available in this field, to find the limitations of these techniques as well as improve upon them. Chapter 1 presents both an overview of generalized techniques at the disposal of researchers looking for links between protein sequence covariance and allostery. The methods most commonly used including mutual information, chemical similarity matrixes, phylogenetic perturbation, and chi-square analysis are reviewed as well as the limits of such approaches to detecting allostery. Chapter 2 explores using a recent phylogenetic correction that has been successful for improving the efficacy of mutual information to predict special contact on the other algorithm types introduced in the first chapter. Chapter 3 is an attempt to detect bias of covariance algorithms on the rigid bodies found in protein structures. Chapter 4 is the description of a novel algorithm, termed COvariance By Sections (COBS), that in many ways is a combination of the methodologies used in Chapter 2 and Chapter 3, whereby we leverage a phylogenetic correction on groups of MSA columns rather than individual columns