Title: The applications of deep learning techniques in protein structure prediction

 
Abstract: Recent advancements in protein three-dimensional (3D) structure prediction using deep learning methods substantially promote biomedical research for improving human health. The high-precision protein structure prediction enables scientists to address many critical questions in biological studies, such as how the protein bindings in host-pathogen interactions cause infectious diseases in healthy individuals. The goal of protein structure prediction using artificial intelligence techniques is to determine the relative positions (i.e., 3D coordinates) of amino acids in three-dimensional space given a known protein sequence. In this talk, Dr. Hou summarizes the key machine-learning-driven approaches for protein structure prediction developed by his research group. He will present how those widely used techniques in the fields of computer vision, natural language processing, and artificial intelligence (i.e., image recognition, language models) are applied and integrated for protein structure modeling. Finally, Dr. Hou will describe the current research focusing on protein folding using computational techniques supporting biomedical research.