Conformational Ensembles from Experimental Data and Computer Simulations

Conformational Ensembles from Experimental Data and Computer Simulations

Poster Abstracts

67-POS Board 27 Efficient Parallel Computation for Flexible Fitting of Cryo-EM Density Map Takaharu Mori 1,2 , Osamu Miyashita 3 , Yuji Sugita 1,2,3 . 1 RIKEN Theor. Mol. Sci. Lab., Wako-shi, Japan, 3 RIKEN AICS, Kobe-shi, Japan. 2 RIKEN iTHES, Wako-shi, Japan, Single-particle cryo-electron microscopy (cryo-EM) is one of the powerful experimental techniques to determine structures of biomolecules at near atomic resolutions. The method does not require crystallization of the sample, and the 3D structure of the target biomolecule is reconstructed from a large number of 2D images. Molecular dynamics (MD) simulations have been often used to construct 3D structures by fitting the all-atom model to low-resolution cryo- EM density map. One of the major problems in such a flexible fitting simulation is that we need large computational costs if we tackle big molecules like ribosome and protein complexes. Recently, we have developed our in-house MD program package (GENESIS), which supports various types of replica exchange molecular dynamics methods (REMD) as well as conventional MD methods for large systems including hundreds million atoms [1]. In this study, we introduced a new REMD algorithm for cryo-EM fitting, where the force constants of the biasing potential are exchanged between a pair of replicas. The method can automatically adjust the strength of the biasing force in each replica to avoid overfitting issues [2]. We also proposed a new parallel computing algorithm for cryo-EM fitting with hybrid MPI/OpenMP schemes, where the simulation system is decomposed into several domains according to the number of atoms in the local spaces. We show performance of GENESIS in cryo-EM fitting simulations for large systems such as membrane proteins and ribosomes. [1] J. Jung*, T. Mori*, C. Kobayashi, Y. Matsunaga, T. Yoda, M. Feig, and Y. Sugita, WIREs Comput. Mol. Sci., 5, 310-323 (2015). [2] O. Miyashita, C. Kobayashi, T. Mori, Y. Sugita and F. Tama, J. Comput. Chem. DOI: 10.1002/jcc.24785

102