Skip to main content
. 2017 Aug 8;73(Pt 9):729–737. doi: 10.1107/S205979831700969X

Figure 3.

Figure 3

(a) AUSPEX plot of I obs versus resolution for PDB entry 4epz. This data set was processed with DIALS and scaled with AIMLESS (Evans & Mushudov, 2013 ▸). For hexagonal ice there should be nine rings visible at this resolution (see Fig. 1 ▸), while for cubic ice there should be three (at 3.67, 2.25 and 1.92 Å). The resolution ranges corresponding to potential ice rings are marked using grey bars (whether hexagonal or cubic ice). Two ice rings are clearly visible at high resolution, while the other (at least one more must be present even for cubic ice at 3.67 Å) was successfully modelled in integration. Hence, when identifying ice rings in integrated data, the presence of all ice rings in question is not a reliable criterion. (b) Background overestimation and underestimation: this enlarged view of the ice ring at 1.918 Å shows the effects of insufficient background correction. The blue line shows the likely background as assumed by the integration program (DIALS). The yellow line shows the likely background caused by ice. The discrepancy causes an underestimation of I obs values to the left and right of the ice ring, resulting in large negative intensity values, and an overestimation of I obs in the ice ring, resulting in a peak in I obs values.