Docking a model into a cryo-EM map with em_placement

Authors

Purpose

The em_placement tool docks the input model in a cryo-EM reconstruction, using the half-maps in an adaptive search strategy.

How em_placement works

The half-maps are compared and analysed to determine signal and noise power as smooth functions in Fourier space. This information is used to predict whether or not a rotation search with a structural model over the full map volume is likely to succeed. If not, the maps are divided into sub-volumes small enough that the noise will be reduced to a level where signal from the rotation search can be detected. Following this, translation searches in the full map volume or the sub-volumes are carried out. Finally, rigid-body refinement is carried out using spherical sub-volumes centered on potential solutions to give the final solutions and confidence indicators.

A log-likelihood-gain (LLG) score greater than 60 suggests that a solution is significantly better than random.

Running em_placement

There is not yet a GUI to run em_placement, which should instead be controlled by an input script using the phil syntax. A basic phil script would look like the following, in which a model derived from the membrane domain of PDB entry 4cof is docked into the reconstruction of the GABA receptor in EMDB entry 11657.

voyager
{
  remove_phasertng_folder = True

  map_model
  {
    half_map = emd_11657_half_map_1.map
    half_map = emd_11657_half_map_2.map
    best_resolution = 1.7
    point_group_symmetry = C5
    sequence_composition = 7a5v.fa
  }

  biological_unit {
    molecule
    {
      molecule_name = 4cofA_membrane
      map_or_model_file = 4cofA_membrane.pdb
      starting_model_vrms = 0.8
    }
  }

}

If the phil script is in a file named docking_script.phil, the search would be run from the command-line as follows.

phenix.voyager.em_placement docking_script.phil

Most parameters specified in this script have been named in a way intended to convey the purpose of that parameter.

The remove_phasertng_folder parameter is activated to clean up the graph database produced by phasertng, which could be used in other circumstances as part of a larger automation framework or for debugging.

The point_group_symmetry feature is only used at the moment to optionally generate a full assembly from a single copy.

The sequence_composition parameter specifies the name of a file containing the sequences of all the components in the reconstruction.

Literature

Read RJ, Millán C, McCoy AJ and Terwilliger TC. Likelihood-based signal and noise analysis for docking of models into cryo-EM maps. To be published.

Millán C, McCoy AJ, Terwilliger TC and Read RJ. Likelihood-based docking of models into cryo-EM maps. To be published.