2019.05.23 Machine Learning approach to link single-molecule FRET and MD simulations for understanding protein
题 目：Machine Learning approach to link single-molecule FRET and MD simulations for understanding protein folding dynamics
报告人：Prof. Yuji Sugita
地 点：北京大学吕志和楼B101报告厅（Rm. B101, Lui Che Woo Building）
Single-molecule (sm) FRET is one of the essential experiments to measure conformational dynamics of proteins in vitro as well as in vivo conditions. SmFRET can detect the distance between donor and acceptor dyes and its dynamics, while the three-dimensional structure information is not able to be obtained. Here, we combined time-series experimental information by smFRET with atomistic MD simulations for refining free-energy landscapes of folding dynamics of WW-domain in water. For the refinement, we utilize a machine learning technique and construct an atomically detailed and experimentally consistent model of protein dynamics. The refined folding model reproduces experimental FRET efficiency and features hairpin 1 in the transition-state ensemble, consistent with mutation experiments. Our machine learning process provides a general framework applicable to investigating conformational transitions in other proteins.
1. Linking time-series of single-molecule experiments with molecular dynamics simulations by machine learning, Yasuhiro Matsunaga, Yuji Sugita, eLife (2018) 7, e32668.
2. Refining Markov State Models for conformational dynamics using ensemble-averaged data and time-series trajectories, Yasuhiro Matsunaga, Yuji Sugita, J. Chem. Phys. 148, 241731 (2018).
3. Sequential data assimilation for single-molecule FRET photon-counting data, Yasuhiro Matsunaga, Akinori Kidera, Yuji Sugita*, J. Chem. Phys. 142, 214115 (2015).
Yuji Sugita received his B.S., M.A. and Ph.D. in Chemistry at Kyoto University. He then worked his postdoc in RIKEN and Institute for Molecular Science separately from 1998 to 2002. After that, he moved back to Kyoto University, the Institute of Molecular and Cellular Biosciences as a lecturer. From 2007, he worked in RIKEN as an associate chief scientist and was promoted to chief scientist in 2012. Research in his lab is focused on computational biophysics, including 1) clarify the dynamics of membrane and membrane proteins; 2) clarify the protein-glycan interactions; 3) develop new theory and computational method that integrates both theoretical chemistry and biophysics; 4) understand macromolecular crowding environments in cells.