May 8, 2017
Jolley Hall, Room 309
"Efficient Geometric Approaches for Mining Protein Structure from Cryo-EM Density Maps"
Adviser: Tao Ju
A protein's 3D structure is the key to understanding its biological function. In recent years, cryo-electron microscopy or cryo-EM has established itself as a mainstream technique to capture proteins' structure at near-native conditions. However the vast majority of cryo-EM data are at medium (5-10A) or low (>10A) resolutions, which is insufficient to capture a protein's atomic structure. Fortunately, at such resolutions, some intrinsic structures of a protein, such as secondary structure elements (SSEs) and smooth C-alpha backbone fragments (SCBFs), can be recognized or robustly detected. In this dissertation, we present efficient protein fitting pipelines to recover a protein's atomic structure given the protein's cryo-EM density map by leveraging the intrinsic structure information detected from the density map. Specifically, we first compute the correspondences between the protein features (SSEs or SCBFs) detected from the cryo-EM density map and those extracted from the template model. Then we fit the template model into the cryo-EM density map guided by the obtained correspondences.