Deep learning-based prediction of electron density maps of proteins

Abstract

Listen

Translate article

Determining the structure of a protein has long been one of the most important endeavors in biology. Among the mainstream methods, X-ray diffraction contributes to more than 85% of the Protein Data Bank. There are two questions that have plagued the X-ray crystallographic community - the rationalization of protein crystallogenesis and the finding of readily usable solutions for the phase problem. In the process of X-ray diffraction, each spot in a diffraction pattern from the X-ray can be considered to have 2 main components - the intensity and the relative phase.