next up previous contents index
Next: OPTIMIZE optimize Up: Optimization of the model Previous: WRITE_SCHEDULE write   Contents   Index

ENERGY -- evaluate MODEL given restraints

VIOL_REPORT_CUT = <real:35> 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 4.5 999 999 999 999 4.5 4.5 4.5 4.5 4.5 4.5 999 6.5 4.5 4.5 4.5 4.5 4.5 999 999 999 4.5 4.5 cutoffs for reporting relative violations
VIOL_REPORT_CUT2 = <real:35> 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
OUTPUT = <string:1> 'LONG' 'SHORT' | 'LONG' | 'VERY_LONG' | 'GRADIENT' | 'SYMMETRY' | 'ENERGY_PROFILE' | 'VIOLATIONS_PROFILE'
NORMALIZE_PROFILE = <logical:1> off whether to normalize energy/violations profiles or not, by the number of terms per residue
SMOOTHING_WINDOW = <integer:1> 3 profiles are smoothed over 2*SW + 1 residues
SCHEDULE_SCALE = <real:35> 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 factors for physical restraint types in scaling the schedule
SCHEDULE_STEP = <integer:1> 1 schedule step for optimization
FILE = <string:1> 'default' partial or complete filename
ASGL_OUTPUT = <logical:1> off whether to write output for ASGL
SCHEDULE_STEP = <integer:1> 1 schedule step for optimization
TOPOLOGY_MODEL = <integer:1> 3 selects topology library: 1-10
RADII_FACTOR = <real:1> 0.82 factor for van der Waals radii
SPHERE_STDV = <real:1> 0.05 standard deviation of soft-sphere repulsion
DYNAMIC_SPHERE = <logical:1> on whether to use dynamic soft-sphere repulsion terms
DYNAMIC_LENNARD = <logical:1> off whether to use dynamic Lennard-Jones energy terms
DYNAMIC_COULOMB = <logical:1> off whether to use dynamic Coulomb energy terms
DYNAMIC_MODELLER = <logical:1> off whether to use dynamic MODELLER non-bonded restraints
DYNAMIC_ACCESS = <logical:1> off whether to use dynamic accessibility energy terms
EXCL_LOCAL = <logical:4> on on on on whether to exclude bonds, angles, dihedrals, explicit excl pairs from the homology-derived distance rsrs
LENNARD_JONES_SWITCH = <real:2> 6.5 7.5 the range for Lennard-Jones interaction smoothing to 0
COULOMB_SWITCH = <real:2> 6.5 7.5 the range for Coulomb interaction smoothing to 0
RELATIVE_DIELECTRIC = <real:1> 1.0 relative dielectric constant
CONTACT_SHELL = <real:1> 4.0 distance cutoff for calculation of the non-bonded pairs list
UPDATE_DYNAMIC = <real:1> 0.39 when to update non-bonded pairs list
NLOGN_USE = <integer:1> 15 number of residues at which to begin using the N Log N non-bonded pairs routine
COVALENT_CYS = <logical:1> off whether to consider SG-SG covalent bond similar to polypeptide chain when proximity of residues along the sequence is considered. If PATCH_SS_MODEL is done, then make it ON.
RESIDUE_SPAN_RANGE = <integer:2> 0 99999 range of residues spanning the allowed distances; for MAKE_RESTRAINTS, PICK_RESTRAINTS, non-bonded dynamic pairs

Output:
MOLPDF

Requirements:
MODEL & restraints

Description:
The main purpose of this command is to compare spatial features of the current MODEL with the selected restraints in order to determine the violations of the molecular pdf. It lists variable amounts of information about the values of the basis, feature, and molecular pdf's for the current MODEL. All arguments that affect the value of the molecular pdf are also relevant for the ENERGY command.

Within this routine only, the scaling factors for the physical restraint types are obtained from the SCHEDULE_STEP step of the current schedule, multipled by SCHEDULE_SCALE (the original values are returned upon exit from the routine). This allows easy reporting of only a selected subset of all restraints.

Most of the output goes to the log file. The output of the ENERGY command has to be examined carefully, at least at the end of the optimization, when the final model is produced. Additional output files, for the ASGL plotting program are created if ASGL_OUTPUT = on (undocumented).

OUTPUT selects various kinds of output information:

VIOL_REPORT_CUT is a vector with one real number for each physical restraint type. A restraint is reported when its `heavy relative violation' is larger than the corresponding cutoff. The heavy relative violation is calculated by finding the global minimum of a feature according to the restraint, taking the difference between the actual feature in the model and this global minimum, and then normalizing the difference by the standard deviation of the global minimum. The `minimal violation' of a restraint is defined as the difference from the local minimum closest to the value of the feature in the model (with the exception of the spline restraints; see next paragraph).

VIOL_REPORT_CUT2 is similar to VIOL_REPORT_CUT, except that it contains cutoffs for restraint `energies', not heavy relative violations.

The meaning of various other reported properties of the violated restraints is briefly described in the log file. Note that for multi-modal restraints that are described by cubic splines (by default, all multimodal homology-derived restraints), only one optimal value is defined, not the local and global minimum as for the multi-modal Gaussian restraints. As a result, the minimal violations and heaviest violations are the same. For interpreting the seriousness of violations, use the following rule of thumb: There should be at most a few small violations (e.g., 4 standard deviations) for all monomodal restraints. In comparative modeling, the monomodal restraints include the stereochemical restraints and distance restraints when only one homologous structure is used. For the multimodal restraints, there are usually many violations reported because the heaviest violations are used in deciding whether or not to report a violation. In comparative modeling, the multimodal restraints include the $\chi_i$ restraints, ($\Phi$, $\Psi$) binormal restraints and distance restraints when more than one template is used. See also Section 1.8, Question 22.

For profiles:

This command calculates residue energies or heavy relative violations, depending on OUTPUT, for all physical restraint types (there are NPHYCNS of them). Relative heavy violations (Table 2.2) are used because only relative violations of different features are comparable. In both cases, the residue sum is the sum over all restraints that have at least one atom in a given residue. The contribution of each restraint is counted exactly once for each residue, without any weighting. Restraints spanning more than one residue contribute equally to all of them. Thus, the sum of residue energies is generally larger than molecular pdf. The command also calculates the sum of the NPHYCNS contributions for each residue and writes all NPHYCNS+1 columns to a file suitable for plotting by ASGL.

If NORMALIZE_PROFILE is on the profile for each residue is normalized by the number of terms applying to each residue.

All the curves are smoothed by the running window averaging method if SMOOTHING_WINDOW is larger than 0: The window is centered on residue $i$ and extends for (SMOOTHING_WINDOW/2) - 1 residues on each side. Thus, SMOOTHING_WINDOW has to be an even number (or it is made such by the program automatically). The only exceptions are the two terminii, where a smaller number of residues are available for smoothing. The relative weight of residue $j$ when calculating the smoothed value at residue $i$ is $({\sf SMOOTHING\_WINDOW}\index{SMOOTHING\_WINDOW@{\sf SMOOTHING\_WINDOW}}/2 - \vert j-i\vert)$.

The energy or the violations profile is written to the fourth column of the MODEL atomic records (atomic isotropic temperature factors for X-ray structures). Note that all the atoms in one residue get the same number. This output is useful for exploring the violations on a graphics terminal.

See description of OPTIMIZE for the other variables.

Example:


# Example for: ENERGY

# This will calculate the stereochemical energy (bonds,
# angles, dihedrals, impropers) for a given model.

READ_TOPOLOGY   FILE = '$(LIB)/top_heav.lib'
READ_PARAMETERS FILE = '$(LIB)/par.lib'
READ_MODEL FILE = '1fas'
SEQUENCE_TO_ALI ATOM_FILES = '1fas', ALIGN_CODES = '1fas'
SEQUENCE_TO_ALI ADD_SEQUENCE = on, ATOM_FILES = ATOM_FILES '1fas.ini', ;
                ALIGN_CODES = ALIGN_CODES '1fas-ini'
GENERATE_TOPOLOGY SEQUENCE = '1fas-ini'
# Must patch disulfides here to calculate the non-bonded
# energy properly. Also, when you use hydrogens, disulfides
# must always be patched so that sulfhydril hydrogens are
# removed from the model.
PATCH RESIDUE_TYPE = DISU, RESIDUE_IDS =  '17'  '39'
PATCH RESIDUE_TYPE = DISU, RESIDUE_IDS =   '3'  '22'
PATCH RESIDUE_TYPE = DISU, RESIDUE_IDS =  '53'  '59'
PATCH RESIDUE_TYPE = DISU, RESIDUE_IDS =  '41'  '52'
TRANSFER_XYZ
BUILD_MODEL INITIALIZE_XYZ = off

MAKE_RESTRAINTS RESTRAINT_TYPE = 'stereo'
ENERGY DYNAMIC_SPHERE = on


next up previous contents index
Next: OPTIMIZE optimize Up: Optimization of the model Previous: WRITE_SCHEDULE write   Contents   Index
Ben Webb 2004-10-04