|
| Python-based Hierarchical ENvironment for Integrated Xtallography |
| Documentation Home |
The AutoMR graphical interface
AutoMR is a simplified frontend to the molecular replacement program Phaser. It is generally the starting point for running MR in Phenix, due to its ease of use and interaction with other programs (such as AutoBuild). For very advanced cases, the Phaser-MR GUI may be used instead. If you have not previously used Phenix and/or Phaser for molecular replacement, we recommend that you start by reading the overview page, which explains some of the concepts and terminology used by Phaser, as well as guidelines for its use. This document only summarizes the function of the AutoMR GUI itself; more detailed information on program behavior may be found in the command-line documentation and Phaser documentation. Preparing for molecular replacementSee the overview page for approximate rules for when MR will be successful. Unless you have a high-sequence-identity model and are relatively certain of the conformation, it is important to prepare the model(s) by running them through the Sculptor utility before running AutoMR. If you want to try to increase sensitivity by using an ensemble of related structures, you should use Ensembler to superpose them in the same orientation, as AutoMR will not do this automatically. Input and configurationIn the Phenix GUI, there are two interfaces available for AutoMR: one for single-component searches, and one for multi-part structures. The former is shown below:  Files may be added by dragging them into the list widget at top of the window. This interface is more intuitive and easier to set up, but only allows a single search model; if multiple PDB files are supplied, these will be treated as parts of one ensemble, not separate components to search for. A reflections file (amplitudes or intensities, plus sigmas) and at least one PDB file are required. A sequence file may be used to specify the composition (it will also be sent to AutoBuild if model-building is requested), or you may enter the molecular weight below. A list of either RMSDs or sequence identities for each model in the ensemble is required. The multi-ensemble version splits the file input into separate controls:  You may use as many ensembles as you want; each of these may contain multiple related PDB files. The composition is entered in a separate tab:  The number of composition entries and the number of ensembles do not need to be identical; you may simply specify an overall molecular weight if you prefer, rather than entering each sequence individually. OutputOnce Phaser starts running, a tab will appear showing search progress and summary output:  At the end of the run, a result tab will appear listing files,  As a general rule, the LLG should be as high as possible (for a very good solution, values above 1000 are common). If the LLG is negative, this essentially means that the model performed as well as would be expected for randomly placed atoms. The final translation function Z-score (TFZ) is also important, and can be found in the logfile or Phaser summary output. A TFZ above 8 usually means that MR was successful. Only two output files will be generated, an MTZ file (MR.1.mtz) containing phases, and a PDB file containing the placed ensembles. If each ensemble consists of only a single model, the PDB file may be directly used for refinement; otherwise, we recommend running AutoBuild first. Although the MTZ file contains F and SIGF values, these are anisotropy-corrected and should not be used for refinement. Clicking any of the graphics viewer buttons will open both files, with the weighted map coefficients generated by Phaser displayed. The maps are usually very noisy (and if a high-resolution cutoff was imposed, this will also be applied), but inspecting the crystal packing may provide additional clues to whether MR was successful. Next stepsThe model and maps obtained by molecular replacement are usually not directly useful for manual rebuilding or ligand placement, since they often are generated at a lower resolution limit and have undergone only minimal rigid-body refinement. Either automated model-building in AutoBuild or full refinement in phenix.refine is usually appropriate, depending on data quality and similarity of the search model(s) to the crystallized protein(s). For refinement, a combination of rigid-body, individual sites, and individual ADPs is usually recommended, except at low resolution where grouped ADPs may be necessary. Simulated annealing may also be very helpful at this stage. For very marginal solutions, if you are confident of the approximate placement of the model(s) but automated rebuilding and/or refinement fails, the MR-Rosetta protocol is able to improve a poor map by extensive optimization of the model with an electron-density target. MR-SADIn situations where neither molecular replacement nor SAD phasing yields high-quality phases independently, the two procedures may be combined by running AutoMR followed by AutoSol. The phases from MR will be used to locate heavy atom sites in anomalous data, and combined with the experimental phases. To perform this in Phenix, all you need is to run AutoMR with your anomalous data, then click the "Run MR-SAD" button in the result tab. This will launch AutoSol with the input reflections and MR phases already loaded. TroubleshootingSee the MR overview or the Phaser manual for further advice. Reference
| |