Hingefind (Tcl) -

an algorithm to investigate domain motions in proteins.

5/30/97

This page contains documentation specifically for the Tcl version.

Content:

1. General Remarks about the Tcl Version
2. The Tcl Source File for VMD
3. Output
4. About the Tcl Code

1. General Remarks about the Tcl Version

The Tcl (John Ousterhout, 1988) version of Hingefind is designed for interactive use with the free molecular graphics package VMD. This version does not maintain the spatial connectivity of the found domains, as implemented in the X-PLOR version. The VMD version has, however, the advantage that it only requires PDB files as input. Thereby, selections of atoms are easier to adapt to the user's needs than with X-PLOR. E.g., conformations of nonidentical proteins can be compared, as long as pairs of atoms in both structures can be matched sequentially.

2. The Tcl Source File for VMD

There is a single file in this directory which contains the algorithm:

hingefind.tcl

Start up VMD and source the Tcl script by typing

source hingefind.tcl

in the VMD command shell and return. The default setup of the script loads two structures of lactoferrin from the PDB databank. To load these structures, type

load

and return. The procedure load now initializes the program, selects the pairs of corresponding atoms (default: alpha carbons) and initializes some global variables. To partition the protein structures with a given tolerance epsilon, type

partition epsilon

and return. Note that the tolerance epsilon must be entered in Angstrom units (this is different from the X-PLOR implementation). The tolerance should be larger than the imprecision of the coordinates in the two structures (local noise level), but smaller than the total rms deviation of the two conformations. The protein now iteratively finds the rigid domains, sorts them by size, and draws tubes in the color of the domain-id, starting at domain-id 0, the largest found domain (note that this is also different from the X-PLOR version, where domains are numbered starting at 1). To determine the effective rotation axis (hinge) and other information about the relative movement of two domains domid1 and domid2, enter

hinge domid1 domid2

and return. Domain domid1 is the reference domain and domain domid2 the moving domain. Both structures are superimposed by the reference domain. The rotation axis is shown as an arrow with it's orientation representing a left-handed rotation about the axis. A "pivot" point can be found in the middle of the arrow which is connected to the COM of the moving domain in both structures to illustrate the rotation angle. The rotation angle and other useful information about the run, the domains, and the accuracy of the rotational fitting can be found in the output (VMD command shell).

An example VMD session would look like this:

source hingefind.tcl
load
partition 2.0
hinge 0 1
hinge 0 2
partition 1.8
hinge 0 1

The user has to edit the procedure load to select different PDB files for comparison, to change the selection of atoms for comparison, or to customize the hingefind computation by resetting some of the global variables. Here is a partial list of variables which may be changed: $maxccounter: The number of maximum cycles of the "converge" loop. In case the algorithm does not converge within the specified number of cycles (this was very rarely observed at extreme tolerances), a warning message is written. $ndomains: The max. number of domains to be found. Recommended: Set very large to yield full partitioning. Small values (2-5) should be used for test runs. $cutdom: Minimum domain size of the found domains.

The initial rendering of the loaded molecules may also be altered in procedure load. By default molecule 1 is shown in purple and codes with colored tubes for the found domains, while molecule 2 remains a solid white line which follows the backbone.

3. Output

The rotation angle and other useful information about the hinge rotation, the domains, and the accuracy of the rotational fitting are written to the VMD command shell. The information includes:

the pivot point of a rotation
the effective rotation axis (hinge)
the effective rotation angle in degrees
the rmsd (least squ.) of the moving domain
the rmsd of the moving domain after hinge rotation
the relative error (see Hingefind home page)
the projection (deviation) angle beta in degrees (see Hingefind home page)

4. About the Tcl code

A programmer can find more information about available VMD commands and routines at the VMD user's guide. All procedures and variables are documented in the program. The datastructures used to store structural information and domain membership are briefly described here:

list of lists of indicies - Used by procedures criterion, superimpose, seed and convergence to store information about selected atomindices in molecules 1 and 2. A list containing a pair of lists of equal length. The first stores the atom indices of molecule 1. The second list stores the corresponding atom indices of molecule 2.
list $dindex - List of domain membership indices for each atom of $full1 (atom selection in procedure load). This list gets updated by procedure update_boundaries once a domain has been found.
list $sup_rmsd - Contains a history of conformance of each atom of $full1 with previously found domains. Procedure update_boundaries updates the domain membership indices only for those atoms which show an improved conformance with the new domain. The conformance of an atom is measured by the separation of the corresponding atoms in molecule 1 and 2 depending on the domain chosen for superposition.
'domindex' arrays - after the protein is fully partitioned into domains, the procedure partition transfers the membership information from the list $dindex to the arrays $domindex1($i) and $domindex2($i), which contain the atom indices of molecules 1 and 2 corresponding to each domain $i. The domindex arrays are then sorted in the proceduresort_n_render to be used for the determination of the effective rotation axis in procedure hinge.