Tool: Fit to Segments

Fit to Segments rigidly fits atomic structures or maps into segmentation regions from Segment Map. Along with that tool, it is part of the Segger package described in:

Quantitative analysis of cryo-EM density map segmentation by watershed and scale-space filtering, and fitting of structures by alignment to regions. Pintilie GD, Zhang J, Goddard TD, Chiu W, Gossard DC. J Struct Biol. 2010 Jun;170(3):427-38.

Comparison of Segger and other methods for segmentation and rigid-body docking of molecular components in cryo-EM density maps. Pintilie G, Chiu W. Biopolymers. 2012 Sep;97(9):742-60.

[back to top: Fit to Segments]

Dialog and Basic Usage

Fit to Segments can be started from the Volume Data section of the Tools menu (more...).

The Structure or Map to fit should be chosen from the pulldown menu of models. It can be a density map, an entire atomic model, or the selected atoms only. If an entire atomic model is chosen, any parts not desired for fitting (such as solvent, ligands, or extra chains) should be deleted from that model.

Clicking the Fit button at the bottom of the dialog performs the fitting. By default, the structure is fit to the selected set of one or more segmentation region(s) using the principal axes method. The regions are made transparent; region transparency/opacity can be adjusted further using the Segment Map shortcuts options or its Regions menu.

Fit information is shown in a table:

• Corr – the correlation, also known as the cross-correlation score, between the map that was segmented (the fitting target) and the map to fit (either specified directly or generated from the atomic structure)
• Atoms – atom inclusion, fraction of atoms within the target map contour; an atom is within the contour if its position in the map has a density value above the current threshold level
• Backbone – backbone-atom inclusion, fraction of backbone atoms within the target map contour
• Clash – fraction of atoms clashing with symmetry copies (if clashes are being evaluated)
• Occupied – density occupancy, fraction of voxels within the target segmentation region(s) occupied by at least one atom of the fit structure
• Molecule – the fit structure, either an atomic model or a map
• Map – the fitting target, the map from which the segmentation region(s) were derived
• Region – the segmentation region(s) used for fitting

Clicking Help shows this page in the this page in the Help Viewer.

[back to top: Fit to Segments]

Options:

• Treat all sub-models as one structure – whether to fit submodels with the same main model number (#2.1, #2.2, etc.) collectively as a single structure instead of separately
  
  
• Density map resolution [r] grid spacing [s] – set parameters for generating a density map from the atomic structure, to allow calculating correlation values for fits. Each atom is described as a 3D Gaussian distribution of width proportional to the resolution r and amplitude proportional to the atomic number (as is done by the command molmap). The grid spacing s is the separation of points along the X, Y, and Z axes of the generated map. The resolution and spacing should be approximately the same as for the map into which the structure is being fit (the map that was segmented). Clicking Fit or Calculate Map will generate the map from the structure if it does not already exist.
  
  
• Which regions to use for fitting:
  • Combined selected regions (default) – use the selected segmentation regions collectively
  • Each selected region – use each selected segmentation region for a separate fit
  • Groups of regions including selected region(s) – generate groups of segmentation regions that include the selected one, try fitting to each group; save the best fits in the table
  • Groups of regions including all regions – combinatorially group subsets of all segmentation regions, try fitting to each group; save the best fits in the table
  
  
• Alignment method:
  • Align principal axes (faster – only 4 fits will be tried) (default) – align centers, then principal axes from longest to shortest. The center and principal axes of a structure are calculated from its atomic coordinates, non-mass-weighted. The center and principal axes of a region or group of regions are calculated from the voxels they enclose; all grid points inside the surface(s) are weighted equally. The principal axes are the eigenvectors of a covarience matrix. An eigenvector gives the direction of each axis, but the signs of these directions are ambiguous. When performing the alignment, the signs are flipped to generate four possible transforms. Only non-reflecting transforms are considered, in which either none or two of the three axes are flipped. The fit with the highest correlation is kept.

    The principal axes of the atomic structure can be shown/hidden with the Fit to Segments menu. The principal axes of segmentation regions can be shown with the Regions menu in Segment Map.

  • Rotational search (try [N] evenly rotated fits) – align centers, then rotate the structure to find the best fit (N orientations tried, default 100). The alignment with the highest correlation is kept.
  
  
• Mask map with region(s) to prevent large drifts (off by default) – whether to mask the map with the selected segmentation regions, zeroing voxels outside the region(s) to prevent large displacements during optimization
• Use Laplacian filter (off by default) – whether to apply Laplacian filtering (as is done by the command volume laplacian) to both the target map and the map to fit (which was either specified directly or generated from the atomic structure). This filter may emphasize contours in images and may help to distinguish the correct fit from incorrect fits at medium-to-low resolutions (~10Å and lower).
• Optimize fits (on by default) – whether to perform a local optimization (as with the command fitmap) of each fit. Optimization uses density both inside and outside the region(s) chosen for fitting (unless the outside density is masked) and can move a structure far from those region(s).
• Cluster fits that are < [d] Angstroms and < [a] degrees apart (on by default) – whether to coalesce fits that are similar in both location (center-of-mass translation < d, default 5.0 Å) and orientation (difference in quaternion angle < a, default 3.0°) into a single fit with their average location and orientation
  Add top [N] fit(s) to list (empty to add all fits to list) – how many fits to add to the list from the current round of fitting (all, if N is not specified)
• Clashes with copies from symmetry [sym] (off by default) – whether to evaluate clashes of the fit structure with its copies by symmetry sym (e.g., C6); clicking Fit or Detect will guess the symmetry of the map (see measure symmetry) if not already specified. The “clashes” reported in the list are the fraction of atoms in the fitted structure within 3 Å of any atom in its symmetry copies. Show displays the symmetry copies. The copies can be removed using the Fit to Segments menu.

[back to top: Fit to Segments]

Fit to Segments Menu

Fit

• Delete selected fits from list – remove the chosen row(s) from the table of fits
• Delete ALL fits from list – empty the table of fits

• Place molecule copies – add a copy of the structure in its fit position (as a new model) for each chosen fit
• Place map copies – if a map was fitted, add a resampled copy of the map in its fit position (as a new model) for each chosen fit
• Close placed copies – close copies of the fit atomic structure and/or map that were added with the “place copies” options above or the button to Show symmetry copies

• Save chosen fit molecules – save a PDB file of the structure in its fit position for each chosen fit (each chosen row in the table of fits)

• Copy fit map on segmented map grid – add a resampled copy of the map currently specified as the Structure or Map to fit (as a new model) for each chosen fit

• Group regions by visible (Molecule) models – group regions in the current segmentation based on which visible structures they overlap the most; Model Panel checkboxes can be used to set visibility
• Group regions by chains in visible (Molecule) models – group regions in the current segmentation based on which chains of the visible structures they overlap the most; Model Panel checkboxes can be used to set visibility

• Show molecule axes – display the principal axes of the atomic structure as arrows
• Hide molecule axes – hide the principal axes of the atomic structure
• Show overlapping regions – display only the regions that overlap the atomic structure in its current position (all regions can be displayed again using the Regions menu in Segment Map)

• Export fit scores – save information in the table of fits to a text file
• Plot fit scores – create a basic plot of the cross-correlation scores in the table of fits (PNG image, colored bars from left to right representing all fits in the order listed)
• Inter-molecule clash scores – report fraction of atoms clashing with symmetry copies (if clashes are being evaluated) to the Log