MS02-P02 3-D analysis of X-ray mesh scans and data collection strategies for macromolecular crystallography Alexander Popov (ESRF, Grenoble, France) Igor Melnikov (ESRF, Grenoble, France) Gleb Bourenkov (EMBL, Hamburg Outstation, Hamburg, Germany)email: apopov@esrf.frData collection procedures at a modern macromolecular crystallography beamlines rely heavily on automated preliminary experiments and data collection strategy software. The standard methods commonly implement a classical beam bathed single-crystal experiment [1, 2]. Further developments of these methods are aiming in automating the design of optimal data collection strategies for complex multi- crystal experiments, sub-crystal diffraction and multi-positional data collection. Three-dimensional information about positions, dimension and relative diffraction strengths of crystals mounted in the sample holder is obtained via several low-dose two-dimensional raster scans at various orientations of the sample holder. Automated preliminary analysis of diffraction images is carried out in real time by the program Dozor, which identifies the presence of a diffraction pattern from a macromolecular crystal, estimates the diffraction signal and produces the list of candidate diffraction spot positions.. Based on the  Dozor output, the program MeshBest [3] automatically determines the areas which belong to individual crystals as well as the areas of crystal overlap.  Latest version of MeshBest analyses multiple mesh scans simultaneously and provides 3D information on the dimensions, centroid  positions and integral diffraction quality of each crystal resolved on the holder. The best achievable result of data collection for each individual crystal is estimated by the program BEST [2], which now uses empirical 3D model of crystal shape delivered by MeshBest. The strategy-optimization method has been extended to take into account the variations in irradiated crystal volume with a spindle rotation. BEST predicts the diffraction intensity at any moment of data collection as a sum of diffraction intensities of crystal voxels taking into account the profile of the incident X-ray beam and the dose absorbed by each voxel. We will present development and applications of the methods which demonstrate feasibility of auotomated approaches and indicate a possibility of significant improvement in data quality in particular for membrane protein crystals grown in mesophase.

[1] Popov, A. N. & Bourenkov, G. P. (2003). Acta Cryst. D59, 1145–1153.

[2] Bourenkov, G. P. & Popov, A. N. (2010). Acta Cryst. D66, 409–419.

[3] Melnikov, I.& Popov, A. N. (2018). Acta Cryst. D74, 355–365.
Keywords: BEST, MeshBest, Dozor