Experimental phasing

From CCP4 wiki

Heavy atoms and their properties[edit]

Heavy atom refinement and phasing programs[edit]

SHARP was the first "product" of Global Phasing, in the mid '90s. It was a spectacular improvement over existing methods and has been described in Maximum-likelihood heavy-atom parameter refinement for multiple isomorphous replacement and multiwavelength anomalous diffraction methods. SHARP has been regularly improved since then, and the legendary 'week-long runs' now complete in a few minutes. Much has been argued about how SHARP implements Maximum Likelihood, and Phaser and BP3 claim to do better in SAD cases. SHARP does do really well for SAD phasing, but the main advantage still is the ability to describe very complicated experiments, involving multiple wavelengths, various crystals of the same derivative and various derivatives (all with or without anomalous data) in a very appropriate manner. All these can also be done through the autoSHARP pipeline as well. The SHARP interface is web based, excellent on-line help, and very good examples.

A CCP4i interface to autoSHARP is also available. It should be noted that this is (besides the old MLPHARE) the only program accessible from CCP4i that permits MAD phasing.

Maximum likelihood phasing in CNS appeared to my recollection shortly after SHARP and has a perfectly decent implementation that has solved many structures and produces excellent maps. It has an excellent web-based interface, good on-line help, and very good example scripts.

SOLVE was the first truly automated experimental phasing pipeline in the second half of the '90s. Its worth mentioning also in this section, since it uses a number of ingenious solutions to the phasing problem, e.g. correlated phasing, that produce phases of excellent quality and can easily be used just for phasing outside its own pipeline.

SHELXE is again mostly used in the context of the SHELX suite and the HKL2MAP pipeline but is an excellent stand alone solution as well. Its peculiarity is the combination of phasing and density modification tricks in a very quick iterative fashion, and it can produce excellent results.

The newest kid in the block, many do not even realize its existence, since the experimental phasing module of PHASER has been overshadowed by the overwhelming success of the synonymous molecular replacement module. It will only work for SAD cases, but it crucially implements an option to combine molecular replacement with SAD phases. A very attractive idea if you ever have a bad search model and bad phases.

BP3 also implements SAD phasing in a statistical framework most similar to PHASER, which might be hardly surprising if one considers that the main authors of each program worked together to develop the theory. BP3 has been showcased to produce better phases than e.g. SHARP in specific SAD cases. BP3 is the main phase refinement engine behind the CRANK suite.

  • MLPHARE

An old program which does not give as good results as the other programs; not even mentioned in article Experimental phasing with CCP4. Will not hurt if used with really good data, and will not help much either. A direct decedent of PHARE and huge improvement of it, was supposedly implementing Maximum Likelihood on macromolecular phasing (ML). It has been widely recognized that this was only a useful approximation. Proper implementation of ML are being used in the programs above.

Automated Pipelines[edit]

  • autoSHARP
  • Solve
  • Phenix.Autobuild
  • AutoRickshaw
  • HKL2MAP