The Z Module is a general facility for carrying out conformational
searches based on rigid-geometry mapping or so-called "zero-order"
minimization. It also includes 1st-order minimization methods (Steepest
Descent and Conjugant Gradient), but the fundamental structure of the method
is nonetheless grid-based.
Reference:
RJ Petrella, M Karplus. "A Versatile Deterministic Method for Conformational
Searches: Application to CheY" (to be published).
* Menu:
* Method:: Z Method
* Syntax:: Syntax of the ZMOD command
* Description:: Description of the various subcommand functions
* Examples:: Usage examples
* Supplementary:: More complex searches
The Z Method
The Z method depends on a partitioning of the conformational space of
a system into subsystems or "subspaces", where a subspace is a subset of all
the degrees of freedom in the system. Once defined, the subspaces can be
searched independently or in combinations. Specifically, the facility searches
all N!/(N-n)!n! combinations of N subspaces taken n at a time, where N and n
are any positive integers, and n<=N. Each grid or combination searches a
set of n subspaces having C(1),C(2),...C(n) conformations across a total of
C(1)*C(2)*...C(n) conformations. For example, if there are 4 subspaces, each
containing 20 conformers, taking 2 subspaces at a time would produce 6 grids
(4!/2!2!), each with 400 grid-points or energy calculations. The overall
search results in a subspace, the "product subspace", that is the union of all
the starting or "reactant" subspaces. Hence, when n > 1, the product subspace
is always larger than each of the reactant subspaces. For each grid, the
subspaces not being searched are held fixed at their initial conformations
(internal coordinate values occurring in the IC table prior to any of the
grid searches having been done). The conformers resulting from the search
may be written to a .dcd ("trajectory") file or to an output conformer file.
The latter is formatted in such a way as to allow it to be used as an input
conformer file in subsequent searches.
In the current implementation, the degrees of freedom are internal
coordinates, although they could, in principle, include Cartesian coordinates
as well. The degrees of freedom--proper or improper dihedral angles, bond
angles, or bond lengths--must be defined. Then, conformers for each subspace
are read from one or more input conformer files, which contain lists of the
degrees of freedom, their possible values, and corresponding subspace numbers.
Hence, the input conformer files define the subspaces as well as specifying
their possible conformations.
Before beginning the search, the N subspaces to be searched must be
"loaded." This loading feature allows for searching various parts of the
system without having to redefine substructures or change the numbering of
subspaces--i.e. all the substructures/subspaces may be defined once and then
subsets can be selected for inclusion in the searches by use of the loading
feature.
A subspace has associated with it an atomic "substructure", which is
comprised of the atoms that are to be included in the energy calculations.
For each overall search, the set of atoms included in the calculation is the
union of the sets of atoms in all of the loaded subspaces. Hence, for a
given subspace combination, or grid, within the larger search, atoms
corresponding to subspaces not being searched are also included in the
calculation. Note that an atom can be (and usually is) a part of multiple
substructures, but in general, a degree of freedom should not be a part
of multiple subspaces.
Subspaces can be searched in multiple regions. For the sake of
simplicity, this is done essentially by replicating subspaces, rather than by
the use of a separate "region" layer in the hierarchical formalism of the
method. This means that rather than assigning a subspace two search regions,
the subspace is simply replicated and the two resulting subspaces are then
assigned to different categories or types. A search can be done over all
N!/M!m!(N-M-m)! combinations of the N subspaces, taken M + m at a time, where
M is the number of subspaces being searched in the first region and m is the
number of subspaces being searched in the 2nd (or "minor") region. This is
discussed separately below under Supplementary.
In an attempt to optimize its utility, the Z module is designed to be
able to be entered and exited easily and often over the course of a single
charmm script. This allows sections of Z module commands in CHARMM scripts
to be conveniently interspersed with other CHARMM commands and functions.
Note, however, that some standard miscellaneous CHARMM commands, such as
the DEFIne, SET, and CALC commands, and the OPENing of units for writing
output, are supported within the Z module, itself.
Distance and degree-of-freedom constraints may be defined, and then
selected ones may be used (loaded) in a given search. For each new
conformation, the loaded distance constraints (if they exist) are checked
against the structure. If the conformer satisfies all of the distance
constraints, then the loaded degree-of-freedom constraints are checked. If
the degree of freedom constraints are all satisfied, then the energy
calculation on that conformer is done.
Z module searches result in changes in the main coordinate set as
well as in the main IC table.
Other Notes:
- At the start of the Z module run, the ZMEMory command must be given
so as to allocate the memory for the run.
- The Z module uses the CHARMM internal coordinate tables. Hence, in
order for the module to function, 1) an IC table must be present, having
either been generated (GENErate ... SETup command) or read in
(READ IC command); and 2) the IC FILL command must be given.
This command fills the IC table with internal coordinate values derived
from the Cartesian coordinates in the main coordinate set (see also
intcor.doc).
- The Z module requires the use of the BYCC list-builder (BYCC keyword in
the NBONDs or UPDAte command).
- Note that the Z module modifies the active atom lists, so that the
NBACtive and BACTive commands (which specify the atoms to be considered
in the non-bonded and bonded energies, respectively) should be invoked
after exiting the module and before calculating CHARMM energies or forces
with non-Z-module commands. The same is true for the fixed atom list:
the CONS FIX command should be invoked if the CFIX keyword has been used
in the Z module searches (i.e. in the ZSEArch command).
- For the Z module to function properly, The CHARMM code must be compiled
with the ACTBOND and ZEROM keywords in the precompiler keyword list file
(build/xxx/pref.dat). The NOBYCC and GENETIC precompiler keywords cannot be
used with the Z module (ZEROM). The GENETIC and ACTBOND keywords are
incompatible.
- The Z module currently supports CMAP, but not for SD or CONJ minimizations.
It does not support the implicit solvation models, (ACE, GB, etc.) except
for EEF1.
It is not currently parallel.
Syntax of the ZMOD command
[ZMOD functions SYNTAX]
ZMOD enter the Z module
END exit the Z module
Subcommands Keywords
ZMEMory NDOFs {int} NSUBspaces {int} NCONFormers {int} NVALues {int}
NATOm {int} CSIZe {int} NDIStances NDAToms NDFConstraints MXLN {int}
ZDEFine DOFR {int} internal coordinate specification
{ATOM SELECTION X 2} REVErse
/
SUBStructure {int} or
\
ALIAs {int}
DIST {int} {ATOM SELECTION X 2} GTHAn {REAL} LTHAn {REAL}
DFCO {int} DOF {int} GTHAn {REAL} LTHAn {REAL}
ZREAd CONF RUNI {int} WUNI {int} MISSing REDUndant SSDIsorder
ZLOAd NULL SUBS {int1 int2 int3 ... } CLEAR
DIST {int1 int2 int3 ... } CLEAR
DFCO {int1 int2 int3 ... } CLEAR
ZSPECifications
ZSEArch TAKE {int} WRUN {int} ECUT {real} VCUT {real}
WCOM MINA SPECifics CFIX MSD NOOR WDUN NDST
SD {int} CONJ {int}
ZMIN LOAD
ZSET
ZRANdom SEED {int} WUNI {int} ECUT {real} VCUT {real}
SUBS {intI1 intJ1 intI2 intJ2 intI3 intJ3...} or
SUBS ALLSubspaces {int}
ZFIL ECUT {real} VCUT {real} VLCUT {real} WUNI {int}
LECUt {intI1 intJ1 intI2 intJ2 intI3 intJ3...} END
ZMOD Subcommand Description
-------------------------------------------------------------------------------
ZMOD
END
Enter (ZMOD) and exit (END) the Z module.
-------------------------------------------------------------------------------
ZMEMory
Allocates memory for the entire module. It and several subcommands must be
specified before any other ZMOD subcommands are given. NDOFs is the number of
degrees of freedom to be defined in the system; NSUBspaces is the number of
subspaces/substructures to be defined; NCONFormers is the total number of
conformers in all subspaces to be read in via conformer files; NVALues is the
cumulative length of all conformer files (total number of degree of freedom
values, one value per line) to be read in; NATOM is the total number of atoms
in all substructures (since atoms can occur in multiple substructures, this
number may exceed the total number of atoms in the system); CSIZe is the
maximum number of degrees of freedom that occur in any one conformer
(optional). NDIST is the number of distance constraints (required only
if distance constraints are being used). NDAToms is the total number of atoms
used in all of the distance constraints (required only for distance
constraints). NDFConstraints is the number of degree-of-freedom constraints
(required only for use of DOF constraints the degrees of freedom (DOFR),
the substructures (SUBS), distance constraints (DIST), and internal coordinate
constraints (DFCO) which must all be numbered serially from 1. The degrees
of freedom (DOFR) syntax is similar to that for editing the IC table
(see IC EDIT), and involves either bond distances (BOND), bond angles (ANGL),
or improper or proper dihedral angles (DIHE). The IC table should have been
filled "IC FILL" prior to the ZDEFine DOFR command. Importantly, when a degree
of freedom (DOFR subcommand) is thus defined, its initial value from the IC table
is stored and used as a default value when the subspace containing that degree
of freedom is included among the N subspaces in a search but is not one of
the n subspaces being searched on the current grid.
Although the degree-of-freedom definitions can be changed easily with the
ZDEFine DOFR command, for convenience, the design of the Z module is
intended to allow the user to define all the necessary degrees of freedom
initially, and then to perform all searches on the basis of that initial set of
definitions.
The MXLN subcommand can be used in parallel execution to increase memory associated
with the handling of data output
-------------------------------------------------------------------------------
ZDEFine
The substructure definition contains two atom selections: the first
defines the atoms that make up the substructure, and the second defines
the moving atoms--i.e. the atoms whose positions are to be initialized and
rebuilt during the searches. By default, the Cartesian coordinates are built
from the internal coordinates in the order of the atom numbers in the
substructure, lowest to highest. The REVErse subcommand causes the substructure
to be built up in the opposite sense, from higher atom number to lower. This
reversal is applied only to those substructures to which the subcommand is
assigned (and their aliases). The "REVErse" subcommand is often necessary, for
example, in building up loop structures from N-terminal and C-terminal halves.
(The alternative to explicitly defining a substructure is to define it
as an alias of another one by use of the subcommand "ALIAs {int}."
This is useful in combination with the ZCATegorize MINOR and ZSEARCh MINOR
commands to search the same subspace in two different regions.
See Searching Subspaces in Two Regions, below.)
The distance constraint definitions also contain two atom selections
each. Each constraint is applied to the distance between the centers of
geometry of the two selections. The distance constraint number
(serially ordered from 1), two atom selections, and upper and lower distance
bounds must be given for each distance constraint.
The internal coordinate or DOF constraint (DFCO) definitions require the
specification of the DOF constraint number as well as the integer corresponding
to the actual degree of freedom as defined in the ZDEFine DOF commands. The
upper and lower bounds for the constraints (in degrees) must also be specified.
The constraints are currently implemented for dihedral, improper, and bond
angles (not bond lengths). The bounds must be in the interval [-360,360]. This
feature is intended for applying constraints to degrees of freedom that are not
explicitly being searched. A good example occurs in loop searches, in which
both halves of the loop are being searched together for low-energy
combinations. The dihedral between the two searched halves of the loop is not
explicitly being searched (it doesn't exist in either half of the loop), but it
may need to be constrained in a simultaneous search of both halves.
ZEDEfine must be invoked once for each DOFR, SUBS, DIST definition.
Example 1:
ZDEFINE DOFR 11 DIHE 44 CA 44 CB 44 CG 44 OD1
This defines the 11th degree of freedom as a dihedral involving atoms CA,
CB, CG, and OD1 of residue 44.
Example 2:
ZDEFine SUBStructure 2 sele resi 20 .around. 15 end -
SELE RESI 20 END
This defines the second substructure as containing all atoms within 15
angstroms of residue 20 and specifies that residue 20 will be rebuilt
in the searches.
Example 3:
ZDEFine DFCO 1 DOF 11 GTHAn -100 LTHAn 100
This defines the first internal coordinate constraint as being
applied to degree of freedom 11 (the dihedral angle as defined above),
with bounds of [-100,100] degrees.
-------------------------------------------------------------------------------
ZREAd
Reads in input conformer files. CONF reads in a conformer file from unit
RUNI and writes the "compressed" conformer file to unit WUNI. The conformer
file is compressed in the sense that there is no redundant information when the
file is read from start to finish--i.e. only the degrees of freedom that change
from one conformer to the next are written. This form of the file, which is
the form stored internally and used by the module, decreases space and memory
requirements and increases speed. ZREAd also sorts each conformer by degree of
freedom internally. ZREAd expects the first conformer of each subspace to be
"complete"--i.e. all its degrees of freedom must be present. Subsequent
conformers in the subspace may be present in compressed form.
The format of the conformer file is as follows (fortran format
I14,I14,I14,F14.7):
1 1 7 18.4857895
1 1 3 180.7483736
1 1 1 50.1098635
1 2 7 24.2836786
1 2 3 180.7483736
1 2 1 50.1098635
1 3 7 30.09877
1 3 3 180.168350
1 3 1 60.87635
2 4 8 64.27840
2 4 9 80.287635
2 4 11 17.981386
2 5 8 56.976233
2 5 9 80.287635
2 5 11 1.8923759
The first column gives the subspace number, the second column gives the
conformer number, the third column gives the degree of freedom, and the fourth
column gives the numerical value taken by the degree of freedom. This example
file has not been compressed. The compressed file would appear as follows:
1 1 1 50.1098635
1 1 3 180.7483736
1 1 7 18.4857895
1 2 7 24.2836786
1 3 1 60.8763500
1 3 3 180.1683500
1 3 7 30.0987700
2 4 8 64.2784000
2 4 9 80.2876350
2 4 11 17.9813860
2 5 8 56.9762330
2 5 11 1.8923759
Note that parts of conformers 2 and 5 have been omitted in the compressed file
because they contain redundant information.
By default, ZREAd expects the input file to have perfectly ordered (and
consecutive) conformer and subspace numbers, beginning with 1. This behavior
can be overridden with the MISSing and SSDisorder subcommand, respectively.
MISSing allows for "gaps" in conformer numbers (they still need to be ordered),
and SSDIsorder allows for disorder in the subspace numbers. This means that,
for example, subspace 2 can occur before subspace 1 in the file. However,
each subspace must be made up of consecutive input lines--i.e. subspaces cannot
be "split" in a given conformer file. ZREAd also expects degrees of freedom
not to occur in multiple subspaces. This can be overridden with the subcommand
REDUndant, which is necessary when subspaces overlap or when the same subspace
is defined twice (either with or without aliasing). If two consecutive
conformers in a conformer file are exactly the same, ZREAd will complain, but
it does not check the entire list for conformer redundancies. ZREAd expects
the first subspace in a file to be subspace 1; this again can be overridden
with the SSDIsorder subcommand.
If multiple conformer files are read in, each requires a separate ZREAd
CONFormer command, and the subspaces of the files will be renumbered
internally, so that they are consecutive. For example, if the last subspace of
the first file is 10, the first subspace of the second file will be called
subspace 11, regardless of the number appearing in the subspace column. In
general, if multiple conformer files are being read, the subspaces should be
well-ordered in all files--i.e. no "gaps" or inverted orders within each given
file. The SSDIsorder subcommand may not help in the case of multiple conformer
files with disordered subspace numbers.
-------------------------------------------------------------------------------
ZLOAd
"Loads" subspaces or constraints to be used in searches. This means the
specified subspaces or constraints are selected out of the sets of all possible
defined subspaces or constraints, and will be used in the searches.
If the "SUBS" subcommand is used, the selected subspace/substructure will be
used in the searches. A subspace/substructure must be loaded in order to be
searched. (Substructure/subspace aliases should also be loaded if they
are to be searched.) Note that if more than one substructure/subspace
is loaded, the "product" subspace/substructure--i.e. the one appearing
in the output conformer file--will generally be larger than each of the
"reactant" subspaces/substructures, since the product is the union of all
the reactant (loaded) subspaces/substructures.
The "NULL" subcommand allows for the loading of empty subspaces--i.e. ones
containing no conformers or ones for which no conformer file has been read in.
This is usefull in cases where, for example, a single subspace corresponds to
(the union of) 2 previously defined substructures. When it is used, the
"NULL" subcommand must occur as the first subcommand in the ZLOAd command.
If the "DIST" subcommand is used, the selected distance constraints will be
applied in the searches.
The "DFCOnstraints" subcommand is analogous to the "DIST" subcommand, except it
is used for loading the internal coordinate (DOF) constraints.
The CLEAr subcommand causes the previously loaded subspaces (or distance
constraints) to be unloaded prior to the parsing of any other subcommand in the
ZLOAd command.
-------------------------------------------------------------------------------
ZSPEcifications
This command causes the currently stored (user defined) search
specifications to be written to standard output.
-------------------------------------------------------------------------------
ZSEArch
Carries out N!/(N-n)!n! grid searches of the N loaded subspaces taken n at
a time, where n <= N. It eliminates conformers that are above specified
thresholds in energy, or outside of distance constraints, and can write
results to a product conformer file or a .dcd file (or both). The output
conformer file is formatted exactly as the input conformer file, except that,
in addition to the subspace, conformer, degree-of-freedom, and d.o.f.
numerical value information, the energies are also written in an additional
column to the right (in 1X,F19.7 fortran format). For convenience, the energy
of each conformer is written next to each of its component degrees of freedom.
If requested, the MSD's (mean square deviations from a target structure) are
written to a sixth column. If writing to a .dcd file is specified, all atoms
in the system are included.
TAKE {int} specifies the number of subspaces to be searched in
combination--i.e. the dimensionality of the grid ('n' in the expression above).
WCOM partially compresses the information in the output conformer file.
For each grid, only the first outputted conformer is written in its entirety;
the remaining conformers include only degrees of freedom that are being varied
in that grid. Since the ZREAd command requires that the first conformer be
"complete," and since the conformers are only accessed serially from the
start of the file, care should be taken when modifying a compressed conformer
output file prior to its use as an input file for a subsequent search. In
general, such modification is not recommended.
ECUT (real} and VCUT {real} are the fixed and variable energy cutoffs,
below which conformers will not be written to the output conformer file. If
both are specified, both cutoffs are used. ECUT is an absolute energy cutoff
value. VCUT is a tolerance above the running energy minimum and hence varies
over the course of the calculation. For example, ECUT 2000 VCUT 30 would
result in the exclusion of all conformers with energies > 2000 kcal/mol OR
>30 kcal/mol above the running minimum (but not necessarily the absolute
minimum).
MINA assigns the structure (main coordinate set) to the global minimum
after the search is completed.
WRUN the unit number to which the output conformer file is to be written.
MWRU is the write unit number to which the minimum-energy conformer file
is to be written.
TAG {int} is the numerical (integer) name given to the product subspace
and written in the first column of the output conformer file.
CFIX subcommand causes the non-bonded energy contributions of fixed-fixed
atom pairs not to be calculated (bonded energy contributions remain).
Fixed atoms are defined as ones that are not specified as moving for any
of the substructures included in a search. I.e. the moving atoms in a search
are taken as the union of the second atom selections of all the ZDEFINE
SUBStructure commands corresponding to the loaded substructures. The fixed
atoms are all the atoms in the system that are not taken as moving.
Since a significant fraction of the atoms included in the searches may not
be moving, the use of CFIX may speed up the calculations without affecting
the relative energies. Note that CFIX will also fix the atoms during
minimization.
MSD calculates the MSDs (mean squared deviations, or squared RMSDs) of the
conformers from the structure in the comparison coordinates. The MSD's are
calculated for all atoms in all the loaded substructures (union of all first
atom selections in the ZDEFine SUBStructure command). Least-squares-fit
reorientation is done prior to rmsd calculation unless the NOORientation
subcommand is also specified. The MSD subcommand currently is not compatible with
WCOMpression.
SPECifics subcommand causes the user-defined specifications that will
be used in the search to be written to standard output.
WDUN unit for writing out structures to .dcd ("trajectory") files.
NDSTeps total number of frames to be written to .dcd [default = 1000]
NOEN causes the energies not to be calculated. This is useful,
for example, when using only distance or internal coordinate constraints,
or when reading in a conformer file only for the purposes of writing out
the corresponding .dcd file.
SD {int} or CONJ {int} cause a minimization of the coordinates to be
carried out at each gridpoint in the search. The coordinates of all
atoms involved in the search (union of all loaded substructures) are minimized,
unless constraints are set prior to the minimization (e.g. CONS FIX can be
used as usual, outside of the Z module). Note that after
each minimization, the unminimized coordinates are restored before the next
point on the grid. This is necessary to preserve the internal coordinates that
are not being explicitly searched but are nontheless affected by minimization.
The non-bonded list is always updated at the start of the minimizations,
provided the update frequency (INBF) is > 0. The energies written to the
conformer file and the structures written to the .dcd file correspond to the
minimized coordinates. The assigned minimum-energy structure (which is
obtained with the MINA subcommand in the ZSEArch command) corresponds to the
unminimized coordinates.
-------------------------------------------------------------------------------
ZMIN
assigns the structure to the minimum indicated. If LOAD subcommand
specified, it assigns the structure to the minimum over all searches done
since the last ZLOAD command.
ZMOD Command Usage Examples
ZMOD !enter module
ZMEMOry NDOF 100 NSUBSPACE 20 NATOM 25000 NCONF 2000 NVAL 10000 !allocate memory
ZDEFINE DOFR 1 DIHE 10 C 11 N 11 CA 11 C
ZDEFINE DOFR 2 DIHE 11 N 11 CA 11 C 12 N
ZDEFINE DOFR 3 DIHE 11 C 12 N 12 CA 12 C
ZDEFINE DOFR 4 DIHE 12 N 12 CA 12 C 13 N
ZDEFINE DOFR 5 DIHE 29 C 30 N 30 CA 30 C
ZDEFINE DOFR 6 DIHE 30 N 30 CA 30 C 31 N
ZDEFINE DOFR 7 DIHE 30 C 31 N 31 CA 31 C
ZDEFINE DOFR 8 DIHE 31 N 31 CA 31 C 32 N
ZDEFINE DOFR 9 DIHE 35 C 36 N 36 CA 36 C
ZDEFINE DOFR 10 DIHE 36 N 36 CA 36 C 37 N
ZDEFine SUBStructure 1 SELE ALL END SELE IRES 1:13 END
ZDEFine SUBStructure 2 SELE ALL END SELE IRES 1:32 END
ZDEFine SUBStructure 3 SELE ALL END SELE IRES 1:38 END
ZDEFine SUBStructure 4 SELE ALL END SELE IRES 1:50 END
open unit 12 write card name echodata
open unit 10 read card name short.conf !conformer file
ZREAD REDU CONF RUNI 10 WUNI 12 !we are allowing redundancy of dofs in the
!conformer file
END ! exit the Z module
!other charmm commands here
ENERGY
MINIMIZE SD
.
.
.
ZMOD !reenter Z module
ZLOAD SUBS CLEAR 2 3 4 !load 3 out of 4 defined subspaces
open unit 15 write card name ener.file
!specify searches that take 2 subspaces at a time
!write to unit 15, assign the minimum to the main coordinate
!set after the search, set product subspace tag to one (for output),
!save (and write) only the conformers within 500 kcal/mol of the running
!minimum energy, write search specifications to standard output
ZSEArch TAKE 2 WRUN 15 MINAssign TAG 1 VCUT 500 SPEC
close unit 15
END !exit Z module
!other charmm input:
COOR RMS ORIENT SELE ALL END
OPEN UNIT 10 WRITE CARD NAME crd/final.crd
WRITE COOR CARD UNIT 10
CLOSE UNIT 10
-------------------------------------------------------------------------------
ZSET
Finalizes or 'sets" the input data in preparation for a search. In parallel
execution, the subcommand also triggers communication of the data to all nodes.
-------------------------------------------------------------------------------
ZRANdom
Randomly selects conformers in one or more subspaces from the set that
has been read in. The SUBS subcommand allows specification of the subspaces
to be included in the random selection and actually triggers the selecion.
It can be followed either by 1) pairs of integers--subspace number followed by the
number of conformers to be selected; or 2) the ALLSubspace subcommand, which
specifies that all subspaces are to be included in the selection, followed
by a single integer, which is the number of conformers to be selected in
each subspace. The subspaces not included in the selection are left as they
are. The SEED (positive integer) subcommand sets the seed for the random
conformer selection. It must be set prior to (in the same ZRANdom command as)
the SUBSpace subcommand. WUNIt is the unit number for writing out the selected
conformers. ECUT and VCUT are fixed and variable energy cutoffs, respectively,
as described under ZSEArch.
examples
ZRAN SEED 189876500
ZRAN SUBS 1 3 2 10 3 15 4 20
--------------
ZRANDOM SEED 189876500 SUBS ALLSubspaces 30 WUNI 19 ECUT -2200
In the top example, the random seed is set, and then random selections of
3, 10, 15, and 20 conformers are invoked for subspaces 1 through 4, respectively.
In the second example, random selections of 30 conformers are are invoked for
all subspaces, with the specified seed and an energy cutoff of -2200 kcal/mol,
writing the output to unit 19
ZFIL
Filters the conformers that have been read in by energy; can set different
energy cutoffs for different subspaces. The LECUt subcommand specifies
the subspaces and their respective fixed energy cutoffs. The LVCUt subcommand
specifies energy bands (tolerance above minimum) local to each subspace,
whereas the VCUT and ECUT subcommands specify global cutoffs, as above.
example
ZFILter LVCUt 5 LECUt 1 50 2 20 3 10 END
Here the fixed energy cutoffs specified are 50, 20, and 10 kcal/mol for
subspaces 1 to 3, respectively. There is, in addition, a cutoff of
5 kcal/mol above the energy minimum for each subspace.
More Complex Searches: Searching Subspaces in Two Regions
In some cases, the user may want to be able to search the same
conformational subspace in two different regions in a series of grids. Take,
for example, the case of 3 rigid helices for which the optimal packing
arrangement is sought. The most efficient search algorithm may be one in
which each helix is allowed to sample a large region of conformational space,
while the other two helices make local adjustments. Hence, each helix would
have two defined regions: one for the large changes and one for the local
adjustments--i.e. a "major" and a "minor" region. As helix A was searched in
its major region, helices B and C would be searched in their minor regions, and
likewise for the other combinations.
Rather than introduce a "region" formalism into the structure of the
Z module--i.e. having each subspace be composed of multiple regions--which
would be complicated for both users and developers, the decision was made to
allow subspaces to have aliases. The alias of a subspace has exactly the same
degrees of freedom and atomic substructure, but it is called separately and can
be assigned different conformations from its primary subspace counterpart
through the conformer file(s). The CATEgorize MINOr command is necessary to
categorize the aliased subspace as "minor", or secondary, so that it is treated
as such in the ZSEARch command. The "MINOR {int}" subcommand must be used in the
ZSEArch command to indicate how many minor subspaces or regions are to be
searched at a time.
ZDEFine SUBSPACE {int} ALIAS {int} is used as an alternative to explicitly
defining an atomic substructure. The integer after ALIAS specifies the number
of the primary subspace to which the alias corresponds. Aliasing is necessary
when searching two regions of the same subspace (i.e. a subspace and its alias)
in a set of searches, because it tells the ZSEArch command that the two loaded
subspaces correspond. This allows the Z module to exclude search grids in
which the same subspace effectivley appears more than once. Without aliasing,
subspaces are combined in searches without regard to whether they are the same
or not, so that results may be redundant or incomplete. The atom selections
and REVErsal specifications are not necessary (and not allowed) when ALIAsing
a subspace.
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Syntax for additional commands in multiple-region searches:
ZCATegorize MINOr CLEAR NONE {int1 int2 int3 ... }
ZSEArch MINO {int} RESS
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Description of additional commands
ZCATegorize
Categorizes a subspace, effectively creating a different search region for
the space. Currently, only one alternative region can be searched for a given
subspace, which is assigned the category (subcommand) "MINOr." The subcommand
"minor" indicates that the subspace is to be included in the set of "MINOr"
or secondary regions that are to be searched by the ZSEARch command (it implies
nothing about size). A set of integers must be specified which refer to the
numbers of the subspaces being categorized as minor. The CLEAR or NONE
subcommands clear any previous categorizations of subspaces as minor.
-------------------------------------------------------------------------------
ZSEARCH MINOR {int} RESS
MINOr {int} specifies the number of subspaces, out of TAKE searched
subspaces, that should be searched in their "minor" or alternative regions
[default=0].,
The RESS ("REpeat SubSpace") subcommand allows multiple versions of the same
subspace, as defined by the ZDEFine ALIAS command, to be searched at the same
time (i.e. on the same search grid). This is not generally recommended and the
user employs this subcommand at his own risk.
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Example
An example of a search using multiple subspace regions, modified from
an example above
ZMOD !enter module
ZMEMOry NDOF 100 NSUBSPACE 20 NATOM 25000 NCONF 2000 NVAL 10000 !allocate memory
ZDEFINE DOFR 1 DIHE 10 C 11 N 11 CA 11 C
ZDEFINE DOFR 2 DIHE 11 N 11 CA 11 C 12 N
ZDEFINE DOFR 3 DIHE 11 C 12 N 12 CA 12 C
ZDEFINE DOFR 4 DIHE 12 N 12 CA 12 C 13 N
ZDEFINE DOFR 5 DIHE 29 C 30 N 30 CA 30 C
ZDEFINE DOFR 6 DIHE 30 N 30 CA 30 C 31 N
ZDEFINE DOFR 7 DIHE 30 C 31 N 31 CA 31 C
ZDEFINE DOFR 8 DIHE 31 N 31 CA 31 C 32 N
ZDEFINE DOFR 9 DIHE 35 C 36 N 36 CA 36 C
ZDEFINE DOFR 10 DIHE 36 N 36 CA 36 C 37 N
ZDEFine SUBStructure 1 SELE ALL END SELE IRES 1:13 END
ZDEFine SUBStructure 2 SELE ALL END SELE IRES 1:32 END
ZDEFine SUBStructure 3 SELE ALL END SELE IRES 1:38 END
ZDEFine SUBStructure 4 SELE ALL END SELE IRES 1:50 END
ZDEFine SUBStructure 5 ALIAs 1 ! 5 is same as 1 !***************
ZDEFine SUBStructure 6 ALIAs 2 ! 6 is same as 2 !***************
ZCAT CLEAR MINOR 5 6 ! categorizing 5 and 6 as minor subspaces (regions) !***
open unit 12 write card name echodata
open unit 10 read card name short.2.conf !conformer file, now contains
!conformers for subspaces 5 and 6 also
ZREAD REDU CONF RUNI 10 WUNI 12 ! allowing redundancy of dofs in conformer file
END ! exit the Z module
!other charmm input
ENERGY
MINIMIZE SD
.
.
.
ZMOD !reenter Z module
ZLOAD SUBS CLEAR 2 3 4 5 6 !load 5 out of 6 defined subspaces !***********
open unit 15 write card name ener.file
!specify searches that take 3 subspaces at a time, one of which is searched
!in its minor region, write to unit 15, assign the minimum to the main
!coordinate set after the search, set product subspace tag to one (for output),
!save (and write) only the conformers within 500 kcal/mol of the running
!minimum energy, write search specifications to standard output
!include minor subcommand *****************
ZSEArch TAKE 3 MINO 1 WRUN 15 MINAssign TAG 1 VCUT 500 SPEC !***************
close unit 15
END !exit Z module
!other charmm input:
COOR RMS ORIENT SELE ALL END
OPEN UNIT 10 WRITE CARD NAME crd/final.crd
WRITE COOR CARD UNIT 10
CLOSE UNIT 10
CHARMM Documentation / Rick_Venable@nih.gov