Multi-objective automatic optimization of crystallography beamline based on NSGA-II

Abstract

BackgroundThe beam quality determined by the beamline status greatly affects the experimental results of macromolecular X-ray crystallography (MX). At present, the optimization of the crystallography beamline of Shanghai Synchrotron Radiation Facility (SSRF) is manually operated by the beamline staffs, which is time-consuming and laborious. X-ray diffractive (XRD) beamline of SSRF realized single-objective automatic optimization of flux based on differential evolution, but this scheme still has certain limitations for MX beamline.PurposeThis study aims to design and implement an automatic optimization procedure based on non-dominated sorting genetic algorithm-II (NSGA-II) on the crystallography beamline system.MethodsFirstly, based on NSGA-II, a multi-objective automatic optimization model of the beamline was established. Then, an automatic optimization procedure was designed and implemented by using Python for the beamline system. Finally, this optimization procedure was tested on BL10U2 of SSRF to optimize the beam flux and position by adjusting the beamline optical components.ResultsThe test results show that the automatic optimization can find the correct optimization objectives Pareto set with two optimized measures of beam flux and beam position within 30 min, and the optimization efficiency is greatly improved when compared with that of manual or previous optimization.ConclusionsNSGA-II based automatic optimization procedure simplifies the optical optimization of beamline and improves the operation efficiency of the crystallography beamlines at SSRF.

Keywords

• macromolecular x-ray crystallography
• nsga-ii
• multi-objective optimization
• pareto set