Talks & Seminars
Title: Computational Analysis of Proteins for Functional Residue Prediction and Structural Neighbor Retrieval
Dr. Ashish Tendulkar (Faculty Candidate), Reliance Industries
Date & Time: October 28, 2013 10:30
Venue: Conference Room, 01st Floor, C Block, Department of Computer Science and Engineering, Kanwal Rekhi Bldg.
Proteins are involved in most cellular functions and hence are one of the most important macromolecules in any given organism. They are made up of amino acids and exist in the form of three dimensional structures. Experimental biologists are interested in uncovering functional residues in proteins in order to understand the mechanism of their function. The search for functional residues usually begins by identifying homologous proteins for the given protein. Structure based homology detection is proven to be the most effective way of identifying related proteins. In this talk, I will present our recent works in (i) Functional residue prediction and (ii) Structural neighbour retrieval for a given protein. In the first part, I will present our recent approach for functional residue prediction in a given three dimensional structure by collective inference. We represent the structure as a weighted, undirected residue interaction network (RIN). Amino acid residues are nodes of a RIN and two interacting residues are connected by an edge. The weight of the edge represents the correlation between the labels of interacting residues. The functional residues are identified by minimizing the combined cost of residue-wise label misclassification and violation of label correlation constraints. We solve this optimization problem in two stages - the first stage minimizes residue-wise label misclassification cost followed by an iterative collective inference stage that adjusts labels predicted in the first stage so as to minimize violations of label correlations. Our approach significantly outperforms state of the art methods on the standard benchmark dataset. This work was published in ACM conference on Bioinformatics, Computational Biology and Biomedical Informatics (ACM-BCB) in 2012. In the second part, I will present our work in structural neighbour retrieval of a given protein. For this task, we model proteins as documents generated from an unknown language. Unfortunately we do not know the true vocabulary of this language. We used the library of fragments or motifs as a surrogate vocabulary and represent each protein in terms of the words in this vocabulary. We then apply latent Dirichlet allocation to infer latent topics arising from the words. We represent and compare proteins in the topic space. This approach outperforms filter-and-refine approaches and perform at par with structure alignment based methods for structural neighbor detection on a standard benchmark dataset. This work was published in Intelligent Systems in Molecular Biology (ISMB) conference in 2011.
Speaker Profile:
Ashish completed his PhD from Department of Computer Science and Engineering (CSE) of IIT Bombay in 2009. He worked with Department of CSE at IIT Madras for two years as a visiting assistant professor and later with Tata Institute of Fundamental Research for a year. Ashish's primary research interest lies in Computational Biology. He is a recipient of Innovative Young Biotechnologist Award from Government of India. He is currently working as a senior data scientist with Reliance Industries Ltd in Mumbai.
List of Talks


Faculty CSE IT
Forgot Password
    [+] Sitemap     Feedback