Professor Daisuke Kihara
Computer Science
Purdue University
West Lafayette, IN

Monday, September 24, 2018
3:00 pm (CT)
2269 Beckman Institute

Abstract

An increasing number of biomolecular structures are determined by cryo-electron microscopy (cryo-EM). The significant progress of the cryo-EM poses a pressing need for software for structural interpretation of EM maps, such as identifying protein main-chains, building all-atom protein models, and model validation. Particularly, tools are needed for maps determined at around 4Å or lower resolution, where finding correct connection of main-chain and assigning the amino acid sequence into EM density map are challenging. Here we report a novel approach we developed for identifying structures from cryo-EM maps. The method named MAINMAST, is a de novo protein structure modeling method that traces a main- chain in an EM map determined at around 4 Å without referring to known protein structures (Terashi & Kihara, Nature Communications, 2018). Compared with two existing programs, MAINMAST showed substantially better performances than the two methods. We also report the MAINMAST performance in the EM Modeling Challenge, a community-wide protein structure modeling experiments from EM maps.