|Python-based Hierarchical ENvironment for Integrated Xtallography|
The phenix.xtriage graphical interface
The configuration of Xtriage is much simpler than most of the other programs in the Phenix GUI; the minimum input is a reflection file and symmetry information, which for MTZ files is automatically extracted. Xtriage is also run automatically by the wizards (AutoSol, AutoBuild, and AutoMR) and the graphical results can be viewed from the GUIs for those programs as well.
To launch Xtriage, open the main Phenix GUI and click on or drag-and-drop a reflection file onto the Xtriage module. The command-line equivalent is phenix.xtriage_gui; most parameters recognized by the command-line program should work here as well. The configuration window is minimal:
The Parameters button opens a window containing the most essential information about the reflection data, including choice of data to analyze if more than one suitable Miller array is present. Xtriage will accept any reflection file format and both intensities and amplitudes, but a Scalepack or MTZ file with separate I+/I- is the most typical input. Most of this information will be extracted automatically if possible; at a minimum, only the input data and crystal symmetry are required. A complete listing of parameters can be found in the manual for the command-line version.
Interpreting Xtriage output
The output of the GUI is nearly identical in content to the output of the command-line version. The analyses performed can be divided into three categories: data strength and completeness, lattice and symmetry properties, and twinning detection. These are shown on three separate tabs in the results window; at the top of each tab, a drop-down menu can be used to scroll to each sub-section. Buttons on the toolbar open this page, the full Xtriage log file, and a dialog for saving individual graphs, respectively.
In general, many of the plots described have a distinctive and easily recognizable shape for strong, problem-free data. In some cases pathologies may be immediately recognized on the basis of abnormal curves (see for instance the section on twinning and pseudotranslation below).
Data quality analyses
Data strength. Xtriage displays two graphs of signal-to-noise ratio, one overall plot by resolution range, and a second combining signal-to-noise and completeness (shown below). These are very similar to the output of data-processing programs such as HKL2000 and the CCP4 suite (MOSFLM/SCALA), and usually do not vary greatly unless the data are systematically incomplete and/or very weak.
This analysis is also used in the automatic determination of the high resolution limit used in the intensity statistics and twin analyses. A separate table shows completeness for low-resolution data alone.
A Wilson plot analysis a la ARP/wARP is carried out, albeit with a slightly different standard curve. This determines overall B-factors (isotropic and anisotropic) for the data. A large spread in the diagonal values indicates anisotropy, as shown below (however, this example, the p9-sad structure included in the examples directory of the Phenix distribution, is actually very good data). The shape of the curve does not vary greatly for most data, and a large deviation from expected values is unusual. However, the resolution at which the maximum of the curve is found will be different depending on the type of molecule crystallized (protein vs. nucleic acid).
A very long list of warnings could indicate a serious problem with your data. Decisions on whether or not the data is useful, should be cut or should thrown away altogether, is not straightforward and falls beyond the scope of Xtriage.
If the input reflection file contains separate intensities for each Friedel mate, a quality measure of the anomalous signal is reported. The p9-sad example, a high-resolution SeMet SAD dataset, is shown again:
Suggested resolution cutoffs at which the anomalous signal disappears are shown by the vertical lines. The p9-sad example is unusually excellent; a more difficult (but still solvable) dataset, the sec17 example in the Phenix distribution, is below:
Xtriage will print a summary of its interpretation of twinning analyses at the top of the tab, including merohedral and pseudo-merohedral twin laws compatible with the lattice. Below is the output for the pseudo-merohedrally twinned porin-twin example data:
Two separate tests for twinning rely on the cumulative distribution of intensities. Both of these result in plots with a distinctive shapes for normal vs. twinned data. The first, the NZ test, is shown below for the untwinned p9-sad data:
Translational NCS (and translational pseudosymmetry) will also result in an abnormal curve, shifted up relative to the expected distribution. Below is the NZ test plot for an untwinned dataset with translational pseudosymmetry:
Because the combination of twinning and translational NCS may result in a curve that appears close to normal, it is important to consider the second test for twinning, the L-test. Here, twinning is indicated by an upwards shift of the curve relative to expected values:
If the crystal symmetry is compatible with one or more twin laws, each of these will be analyzed separately to determine the expected twin fraction. Most of these will result in a very low twin fraction whether or not twinning is present, but the actual twin law, if any, will usually have a significantly higher value, reflected in the shape of the curves. Results from an untwinned crystal typically look like this: