The sensitivity of the EMC algorithm to the light intensity and amount of diffraction patterns in diffraction experiments

Abstract: To understand the function of macromolecules like proteins it helps to know the structure of the molecule. Coherent diffraction imaging is an emerging method that might be used to figure out the structures of macromolecules. In this method diffraction patterns of the macromolecule are recorded by shining light on the molecule from many unknown orientations and detecting the pattern of the diffracted photons. By assembling the diffraction patterns in a specific way and finding the phase of the photons that gave rise to the diffraction patterns, it is theoretically possible to obtain the electronstructure of the molecule and thus the molecular structure. The assembling of several thousand diffraction patterns representing unknown orientations of the molecule is hard to do by hand, but there are several methods that can be used. The EMC (Expand-Maximize-Compress) algorithm is one of those methods. It is an iterative algorithm that tries to create a model describing the Fourier Transform of the electron density of the molecule by maximizing each diffraction patterns fit to the model. This work examines how sensitive the EMC algorithm is to datasets with few diffraction patterns or a low intensity of the light being diffracted by the molecule, for the proteins phytochrome and lysozyme. The result of the work could be used to make sure enough data in collected in real experiments. Diffraction patterns simulated with the program Condor is used in this work, instead of diffraction patterns from real experiments.EMC finds the correct model when the data set contains about 1/3 fewer photons for the smaller more symmetrical molecule lysozyme than it does for phytochrome. This might be because the shapes in lysozymes diffraction patterns are larger than in phyochrome’s patterns. For phytochrome the EMC algorithm assembled the diffraction patterns correctly, with fewest photons for the light intensity 0.764 J/μm2 and 1250 diffraction patterns. For lysozyme it was with an intensity 1.910 J/μm2 and 1425 diffraction patterns. More investigation of the data is needed to understand what factors that affect the EMC algorithms ability to assemble the diffraction patterns correctly.