SAD using CCP4i

From Phaserwiki
Revision as of 11:36, 20 November 2018 by Randy (talk | contribs) (Describe the option of running ARP/wARP instead of buccaneer, which gives substantially better results in this case.)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Data for this tutorial are found here

Reflection data: iod_scala-unique.mtz
Heavy atom sites: iod_hyss_consensus_model.pdb
Sequence file: hewl.pir

This tutorial demonstrates SAD phasing using the Phaser pipeline.

Lysozyme will readily crystallise from a wide range of conditions and yield several crystal forms. One of those, tetragonal lysozyme, is particularly well suited for halide soaking, since it grows from a high concentration of sodium chloride. A dataset has been collected from a lysozyme crystal soaked in 0.5 M potassium iodide. A strong anomalous signal has been detected, and locations of anomalous scatterers have already been found, but you can choose optionally to look for them as part of this run.

  1. Start the ccp4 GUI by typing ccp4i at the command line.
  2. Make a new project called "phaser_tute" using the Directories&ProjectDir button on the RHS of the GUI. Set the "Project" to "phaser_tute" and "uses directory" to the directory where the files for this tutorial are located, and make this the "Project for this session of the CCP4Interface". You will then be able to go directly to this directory in the GUI using the pull-down menu that appears before every file selection.
  3. Go to the Experimental Phasing module, in the yellow pull-down on the LHS of the GUI
  4. Open the Automated Search & Phasing tab, then bring up the GUI for the Phaser SAD pipeline
  5. You can choose (near the top) whether to use SHELXD (if you have the SHELX package installed) or HYSS (if you have the Phenix package installed) to find the substructure before running Phaser. If not, you can use a predetermined substructure, such as the PDB file included with this tutorial.
  6. You can also choose whether or not to run Parrot to carry out density modification and Buccaneer to carry out a very preliminary model building, in the pipeline after Phaser.
  7. All the yellow boxes need to be filled in.
    • Make sure that "Mode for experimental phasing" is set to Single-wavelength anomalous dispersion (SAD). This is the default.
    • Note that Phaser requires F(+), SIGF(+), F(-) and SIGF(-) and not the FP and DANO used by some other programs.
    • If you are using SHELXD or HYSS to determine the initial substructure, a good number of I atoms to ask for would be 8.
    • It is also a good idea to fill in the TITLE.
  8. When you have entered all the information, run Phaser.
  9. Look at the "Final phasing statistics" table at the end of the logfile.
    • There is some explanation on important logfile sections in the Phaser Wiki
  10. Look through the log file and identify the workflow. How many cycles did Phaser need to reach convergence? What are the convergence criteria? How many iodides were in the initial and final substructures? How many were added? Were any deleted?
  11. If you ran the whole pipeline, look at whether one hand gave better statistics than the other in the Parrot and Buccaneer steps. Look at the better Buccaneer model in the corresponding Parrot map, using coot.
  12. Only a quick build is carried out with Buccaneer. If the result is substantially better with one hand than the other, set up a job to carry on building from that Buccaneer model. You could use Buccaneer again, in which case you should open the Model Building module, choose "Buccaneer - autobuild/refine", start from experimental phases, provide the Buccaneer model from the Phaser pipeline, the corresponding Parrot MTZ file and the sequence. Alternatively, you could use ARP/wARP Classic, in which case you should choose the same model but use the MTZ file produced for the correct hand by Phaser, set Refmac to use the Phased ML target and Rfree flags (both in "Required parameters") and make sure the HLA/B/C/D coefficients have been selected. Five cycles of model-building (25 cycles in total) should be enough.
  13. As an alternative, you could choose not to use Buccaneer in the original pipeline run, but rather to follow that by a run with ARP/wARP Classic, choosing the hand with better statistics in the parrot step. When running ARP/wARP, select the option to use a different (pre-weighted) map and choose the parrot map coefficients (parrot.F_phi.F and parrot.F_phi.phi). Select the options to use the free R flag in Refmac, and to use the Phased ML target with the Hendrickson-Lattman coefficients from Phaser (i.e. HLA, HLB, HLC and HLD). Five cycles of model-building should again be enough.