journal menu
![[Figure 7]](qi5003fig7mag.jpg)
|
|
Figure 7
Resolution prediction for JUNGFRAU 16M data collected at the SwissFEL light source. All color bars are in square-root-photon units. (a) A quadrant of the JUNGFRAU 16M image (512 × 512 pixels) with the highest predicted resolution (1.3 Å). (b) The same as (a) but zoomed in to the outer corner of the image, showing high-resolution Bragg peaks. (c) CC1/2 versus resolution for three different merges. Images were sorted according to Resonet predicted resolution and then divided into three groups called `high', `mid' and `low'. The high, mid and low groups included images whose Resonet predicted resolutions lay in the ranges 1.3–2.5, 2.0–2.9 and 2.5–5.2 Å, respectively. The images in each group were processed with dials.stills_process and merged with cctbx.xfel.merge, resulting in three CC1/2 curves. As shown, images with higher resolutions yielded better CC1/2 statistics and indicated that the model can be used to accurately sort images based on resolution. (d) Histogram of the resolutions predicted for all 9592 images containing crystals. I–IV show maximum composite images. A maximum composite image is an image whose pixel value is the maximum across a subset of images (for a further description, see Brewster et al., 2019  ). In this case, the subsets are those images whose resolutions fell within the shaded regions in (d). I, II, III and IV correspond to the intervals 1.3–1.6, 2.0–2.2, 2.7–2.8 and 3.8–5.7 Å and contain 640, 640, 639 and 639 images, respectively. The average resolution in each maximum composite image is labeled by a black dashed circle. The large peaks in these images are from salt crystals or other parasitic scatterers in the beam, and the smaller, more densely packed peaks represent Bragg reflections. |
![[Figure 7]](qi5003fig7.jpg)
 | STRUCTURAL BIOLOGY |
ISSN: 2059-7983
access
