The emplace_local tool docks the input model within a sphere of density centered on a defined point in the reconstruction, returning the docked model, a local weighted map, and quality of fit statistics.
Ideally, half-maps are provided, then the chosen region of density is analysed for local signal and noise (both of which are resolution and direction-dependent) to define a likelihood target for rotation and translation searches, followed by rigid-body refinement.
If half-maps are unavailable, just a full map can be provided and then the likelihood target is defined by assuming that signal-to-noise drops off isotropically to the stated resolution limit.
The quality of fit can be judged by three factors:
1) whether there is a unique solution; 2) the log-likelihood-gain (LLG) score, which should preferably be greater than 60; 3) the map correlation coefficient to the weighted map used for docking.
The basic inputs are:
model file two unmasked half-map files (preferred) or full-map file nominal resolution (optional if half-maps provided, but recommended) specification of center of search region
Note that the search center can be specified explicitly with the sphere_center keyword (the usual approach) or by translating the search model to the desired search region and not providing a sphere_center keyword.
It is also possible to provide a fixed model as prior information.
phenix.voyager.emplace_local map1=half1.map map2=half2.map d_min=3.3 model_file=model.pdb sphere_center=100,140,170