Design and Development of Deep Learning Based Algorithms for Protein Secondary Structure Assignment and Protein Contact Map Prediction

Abstract

Proteins are among nature s most essential macromolecules that accomplish a range newlineof functions in a living cell, including structural, mechanical, biochemical, and cell newlinesignalling. The ability of a protein to fold into different shapes or conformations is newlineresponsible for the viability of several biological processes in living organisms. It will newlinetake months to years of painstaking effort to experimentally determine the structure of newlinea single protein, and this substantiates the use of computational approaches in protein newlinestructure prediction. These structures are crucial to biologists as they are required newlinefor structure-based drug discovery, predicting protein binding sites and understanding newlineprotein functionalities. When analysing protein structures, it is necessary to visualise newlinethem in a more comprehensible way. Protein secondary structure assignment offers a newlinesimplified and meaningful representation of protein 3D structures. newlineAlthough countless efforts have been made, research in this field remains a daunting newlinechallenge due to the inherent difficulties in interpreting protein structures and understanding the underlying folding principles. Deep learning techniques, a branch of newlineArtificial Intelligence (AI), are often well-suited here, particularly in predicting and newlineunderstanding protein structure and function. These algorithms benefit significantly newlinefrom the vast amount of protein repository data that has been accumulated in recent newlineyears. Deep learning techniques extract information from these data and use them for newlineprediction and classification tasks. This research aims at developing deep learningbased methods for protein secondary structure assignment and contact map prediction newlineproblems.The initial phase of the research is focused on developing deep learning-based newlinemodels for the protein secondary structure assignment problem. Accurate and reliable newlinesecondary structure assignment enriches the structural and functional understanding of newlineproteins.