Figure 4. Data processing strategy for XFEL diffraction data of the Syt1–SNARE complex.
Figure 4—figure supplement 1. Improved crystal lattice refinement with DIALS.
Section of an XFEL diffraction image shown with overlaid integration prediction obtained with the LABELIT method implemented in cctbx.xfel (Hattne et al., 2014) and B) DIALS methods (Waterman et al., 2016) recently implemented in cctbx.xfel. A single reflection with incorrect (A, inset) and correct (B, inset) corresponding integration prediction was selected to illustrate how the new crystal lattice refinement approach implemented in DIALS results in a significant improvement of the integrated intensities. Initial spot-finding results are marked as clusters of red squares, integration masks as clusters of teal squares, and pixels used for background calculation as clusters of yellow squares.
Figure 4—figure supplement 2. L-tests for merged diffraction data sets.
The L-tests were performed using phenix.xtriage (Zwart et al., 2005) for the (A) synchrotron data set, (B) observed XFEL data set with , (C) observed XFEL data set with , (D) simulated XFEL data set with and (E) simulated XFEL data set with . All L-tests are shown for the data sets at their respective limiting resolutions (corresponding to Table 1).
Figure 4—figure supplement 3. Distribution of the optimal spot-finding parameter combinations over all successfully integrated diffraction images.
Each diffraction image was indexed and integrated using every possible combination of the two spot-finding parameters; only the selected optimal integration results (one per each image) were used to construct the heat map. The color of each square represents the number of optimally integrated diffraction images (shown inside each square) for this combination of spot-finding parameters. Parameter distributions for A) experimental XFEL and B) simulated XFEL diffraction data are shown (corresponding to Table 1, columns A, C). Note the much wider distribution in experimental vs. simulated XFEL diffraction data, illustrating the image-to-image variability of serial XFEL diffraction data sets, which necessitated the grid search approach.
Figure 4—figure supplement 4. Inspection of refined direct beam coordinates indicates possible mis-indexed diffraction images.
Direct beam coordinates obtained from (A) diffraction images integrated using default direct beam coordinate refinement radius (4.0 mm) and (B) diffraction images integrated using a restricted default direct beam coordinate search area (0.5 mm). Note the bimodal distribution of direct beam coordinate sets (filled grey circles) clustered around the median (filled yellow circle) and outlier (filled red circles) clustered within a distance (~1.4 mm) approximately corresponding to the length of the c-axis (~292 Å, open blue circle), which indicates incorrect assignment of Miller indices (mis-indexing).