Localized Methods for Protein Interaction Prediction

Identification of protein-protein interactions is important for drug design and the treatment of diseases. We propose a novel threading algorithm, LTHREADER, which generates accurate local sequence-structure alignments and integrates various statistical scores and experimental binding data to predict interactions. LTHREADER uses a profile of secondary structure and solvent accessibility predictions with residue contact maps to guide and constrain alignments. Using a decision tree classifier and low-throughput experimental data for training, it combines information inferred from statistical interaction potentials, energy functions, correlated mutations and conserved residue pairs to predict likely interactions. The significance of predicted interactions is evaluated using the scores for randomized binding surfaces within each family. We first apply our method to cytokines, which play a central role in the development of many diseases including cancer and inflammatory and autoimmune disorders. We tested our approach on two representative families from different structural classes (all-alpha and all-beta proteins) of cytokines. In comparison with the state-of-the-art threader RAPTOR, LTHREADER generates on average 20% more accurate alignments of interacting residues and shows dramatic improvement in prediction accuracy over existing methods. To further improve alignment accuracy for all PPI families, we also introduce the program CMAPi, a twodimensional dynamic programming algorithm that, given a pair of protein complexes, optimally aligns the contact maps of their interfaces. We demonstrate the efficacy of our algorithm on complexes from PPI families listed in the SCOPPI database and from highly divergent cytokine families. In comparison to existing techniques, CMAPi generates more accurate alignments of interacting residues within families of interacting proteins, especially for sequences with low similarity. Thesis Supervisor: Bonnie Berger Title: Professor of Applied Mathematics Acknowledgements First and foremost, I would like to express my deepest gratitude to Bonnie Berger. I consider myself extremely fortunate to have her as my advisor. Her advice and support, both academic and personal, have helped me deal some very difficult times during this decade long journey that began when I worked with her as a UROP student. Her neverending enthusiasm and belief not just in me, but all of her students is something that I will always cherish. I would also especially like to thank Jadwiga Bienkowska, who first inspired the research described in this thesis. I could not have asked for a more knowledgeable and enthusiastic collaborator. Her deep understanding of the field and creative problem solving ability helped us deal with many difficult challenges along the way. I also appreciate the amount of time and effort she put into this research even while working fulltime in industry. I want to thank Jinbo Xu and Rohit Singh for many helpful discussions along the way. Their vast knowledge and experience is far superior to mine. Also, I want to thanks all the members of the computational biology group for their thoughtful feedback during group meetings. Finally, above all, I want to thank my family for their constant love and support throughout my life. I owe everything to my parents who sacrificed so much to give me and my brother the opportunity to be where we are today. This thesis is as much for them as it is for me. And although my father is no longer here, I know he is sharing in my joy. He has been my constant source of inspiration throughout. This thesis was supported by Grant Number 1R01GM081871-01Al from NIH. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of NIH. Table of