ChangeLog.c28
| | |
|
CHARMM 28 |
| c28a1 |
Developmental |
August 15, 1999 |
| c28a2 |
Developmental |
February 15, 2000 |
| c28a3 |
Developmental |
August 15, 2000 |
| c28a4 |
Developmental |
February 15, 2001 |
| c28b1 |
Release |
August 15, 2001 |
|
| | |
|
About CHARMM28 Development |
|
|
|
As an on-going project, CHARMM development will be carried out with
CHARMM version 28 series. The base revision c28a1 is the same as c27b1.
The following describes new features and enhancements incorporated in
the c28 development period.
The following are introduced and modified during the development period
of August 15, 1999 through February 15, 2000.
1. University of Maryland Enhancements
Name : I-Jen Chen and Alexander D. MacKerell, Jr.
Email Address : ijen@outerbanks.umaryland.edu,
alex@outerbanks.umaryland.edu
Institution : University of Maryland, Baltimore
Date : Dec 31, 1999
1.1.1 MMFF Enhancements
MMFF module in CHARMM is enhanced to accept molecules in the SYBYL
MOL2 format. Molecules of interest can be read in as an integral part of
a CHARMM input file (see test/cmmfftest/mmff_mol2.inp and MMFF.DOC USAGE 3c)
or as an external MOL2 database file (see test/cmmfftest/mmff_db_1.inp
and MMFF.DOC USAGE 3d).
Files Modified:
source/fcm/exfunc.fcm
/io/mainio.src
/mmff/merckio.src
/mmff.src
/util/string.src
New Testcases:
test/cmmfftest/mmff_mol2.inp
/mmff_db_1.inp
/data/mmff_db.mol2
Modified documentation
doc/mmff.doc
..............................................................................
1.1.2 SOLANA: Rotational Correlation Time of Water
Calculation of rotational correlation times corresponding to the three
rotational motions of a water molecule has been added to the solvent
analysis code. The three rotational motions refer to motion around the
dipole axis (twist), around an axis perpendicular to the molecular
plane (rock) and around an axis parallel to the H-H vector (wag) (Ref 1).
The correlation time is calculated by fitting the exponentional decay part
of the corresponding time correlation function C(t) to an exponentional
function of the form C(t) = A exp(-t/tau) where tau is the correlation time.
The direct correlation functions were calculated via direct method using the
CORFUNC subroutine in the CORREL.SRC. The calculation can be invoked by
assigning a non-zero integeer value to the keyword RCOR. For the details
of input data please see CORMAN.DOC.
REFERENCES:
1. Johannesson, H. and Halle, B. J. Am. Chem. Soc. 1998, 120, 6859-6870
2. Wallqvist, A. and Berne, B. J. J. Phys. Chem. 1993, 97, 13841-13851
Files Modified:
source/correl/solana.src
Testcase Modified:
test/c27test/solanal2.inp
..............................................................................
1.1.3 SOLANA: Hydration Number Calculation
Calculation of hydration number or, in general, the number of solvent
molecules within a specified distance of a multi atom or single atom site
has been added to the solvent analysis code. This method counts the number
of solvent molecules within RHYD and included normalization for the number
of site atoms and time frames. Since this is done within the solvent-site
g(r) analysis routine keywords such as ISDIST, MGN, DR and RSPH should be
specified. Hydration number calculation can be invoked by specifying
a non-zero integer value to the keyword IHYDN.
Files Modified:
source/correl/solana.src
Testcase Modified:
test/c27test/solanal2.inp
------------------------------------------------------------------------------
1.2 Further Enhancements to the Monte Carlo Module
Name : Aaron R. Dinner
Email Address : aaron@bellatrix.pcl.ox.ac.uk
Institution : University of Oxford
Date : December 9, 1999
1) Steepest descents minimization prior to application of the
acceptance criterion is now allowed.
2) Separate temperatures for each move group are now allowed.
3) NOE constraint terms are now supported.
4) The Tsallis (generalized) acceptance criterion is now allowed.
5) A minor "bugfix" regarding the Delta-E column in the energy table
printed during MC: previously, when the running total of the energy
and the value obtained upon checking the energy were exactly the
same, the value of the energy was printed rather than their difference
(zero).
Files Added:
source/mc/mcmini.src
Files Modified:
source/mc/mc.src
/mcener.src
/moveio.src
/movead.src
/mvdihe.src
/fcm/mc.fcm
/energy/printe.src
/charmm/iniall.src
/nbonds/enbond.src
/enbondg.src
/enbfast.src
Modified Documentation:
doc/mc.doc
------------------------------------------------------------------------------
1.3 CHARMMRATE: CHARMM - POLYRATE Interface
Name : Jiali Gao, C. Alhambra, J. Corchado, M. L. Sanchez,
J. Villa, and D. G. Truhlar
Email Address : gao@chem.umn.edu
Institution : University of Minnesota
Date : December 29, 1999
CHARMMRATE is a module of CHARMM for interfacing it with POLYRATE.
The polyrate main program becomes a subprogram of CHARMM. POLYRATE can be
called to carry out variational transition state theory calculations with
multidimensional semiclassical tunneling contributions. When POLYRATE needs
the value or gradient of the potential energy surface, it calls a set of
interface routines called hooks. The hooks in turn call CHARMM routines
for energies and gradients calculated by molecular mechanics or QM/MM methods.
CHARMMRATE-version 1.0 is a module of charmm-version 28 for interfacing
it with POLYRATE-version 8.1.1. CHARMMRATE-version 1.0 will be distributed
as part of the CHARMM program, beginning with version 28 of CHARMM. The
user will also require the CRATE utility for modifying POLYRATE to make it
compatible with CHARMM. The prospective user of CHARMMRATE should obtain
a valid license for CHARMM and valid licenses for POLYRATE and CRATE from
the University of Minnesota (http://comp.chem.umn.edu).
Files Added:
source/prate/charmmrate.src
/fcm/charmmrate.fcm
build/UNX/prate.mk
Files Modified:
source/charmm/charmm_main.src
/install.com
New Documentation: !! documentation and testcases have not
doc/charmmrate.doc !! been completed as of Feb. 10, 2000
!! should be done before c28a3
New Testcases:
test/c28test/cr01.inp
/cr02.inp
test/data/cr01.top
/cr01.par
/cr02.top
/cr02.par
------------------------------------------------------------------------------
1.4 ACE Enhancement
Name : Michael Schaefer
Email Address : schaefer@brel.u-strasbg.fr
Institution : Univ. Louis Pasteur (Strasbourg) & Harvard
Date : Aug 12, 1999
(1) ACE can be used with BLOCK/REPLICA (multiple copies of part of the system)
(2) atoms can be fixed (CONS FIX) when using ACE
(3) the hydrophobic solvation term of ACE is no longer included in the ELEC
energy term (as it should be, since it is non-electrostatic);
WARNING: if your analysis of CHARMM log files relies on this, you have
to modify your approach!
(4) the parameters of the self energy potential of ACE (1/omega, sigma, mue)
have been moved to the HEAP, as suggested by B. Broux; this reduces the
use of permanent memory significantly, if you compile CHARMM with ACE,
but don't use ACE in an application
(5) warning messages when using ACE for atom types whose volume has not been
assigned in the parameter file; so if you use ACE with conventional
param19.inp (w/o volumes), there are lots of warnings; if there are
just a few warnings, you have to check whether the atom types concerned
are present in the system
(6) capability of CHARMM compiled w/o ACE to read parameter files with the
volume table added -- this avoids unwanted error messages
(7) the HEAP space used by ACE is freed when switching back from ACE to
another electrostatic potential (e.g., RDIE)
(8) included volume spcification for TIP3P atom types and some comments
in the parameter file ./test/data/param19-1.2.inp
(9) improved testcase (./test/c27test/ace1.inp) and updated documentation
in ./doc/ace.doc
Files involved:
source/charmm/iniall.src
/energy/eutil.src
/printe.src
/fcm/energy.fcm
/inbnd.fcm
/io/parmio.src
/manip/cstran.src
/nbonds/ace.src
/acean.src
/enbond.src
/nbonds.src
/nbutil.src
test/c27test/ace1.inp
/data/param19-1.2.inp
doc/ace.doc
------------------------------------------------------------------------------
1.5 BYCC Modification
Name : Robert J. Petrella
Email Address : petrella@tammy.harvard.edu
Institution : Harvard University
Date : December 31, 1999
The only modification involves the removal of the need for cluster files.
The algorithm in SUBROUTINE CLSBYRULE clusters the atoms automatically when
the MKCLuster (make clusters) command is given. Hence, all the user must do
to implement the BYCC algorithm is (1) give the command MKCLusters and
(2) specify BYCC in the NBOND command
Files Modified:
source/nbonds/nbutil.src MKCLUST shortened and CLSBYRULE added
/io/rtfio.src RDCLUS removed
/mainio.src RDCLUS removed
doc/nbonds.doc
test/c27test/testbycc1.inp
/testbycc2.inp
------------------------------------------------------------------------------
1.6 TSRI Fixes and Enhancements
Name : Charles L. Brooks, III, Michael F. Crowley
Email Address : brooks@scripps.edu, crowley@scripps.edu
Institution : The Scripps Research Institute
Date : January 15, 2000
1.6.1 Compilation Enhancement
MPI installation procedures are reworked to modify machine specific
Makefiles. Makefile_gnu is modified for pgf compiler and g77 compilation
options are added to increase performace by ~100%. The optimization level
is altered in nbondg.src compilation for Linux PGI compiler.
Files modified:
install.com
build/UNX/Makefile_cspp
/Makefile_gnu
/nbonds.mk
1.6.2 Documentation Update
Rg/RMSD restraint and Distance Matrix restraint are documnented.
File Modified:
doc/cons.doc
1.6.3 Memory Save and Bugfixes
plnod0 (prnlev node 0) is added and broadcasted to other nodes.
Changes are for saving memory, readcv bugfix, initializing variables.
A bug is fixed in stream file handling.
Files Modified:
source/charmm/miscom.src
/correl/anacor.src
/solana.src
/dynamc/dcntrl.src
/dynio.src
/dynsub.src
/trnphi.src
/energy/eintern.src
/fcm/heap.fcm
/parallel.fcm
/gener/genpsf.src
/image/nbndgcm.src
/io/psfres.src
/rtfio.src
/trajio.src
/machdep/paral1.src
/manip/corman2.src
/cstran.src
/dynanal.src
/hbanal.src
/intcor.src
/minmiz/abner.src
/misc/hbuild.src (Milan's parallel bugfix)
/nmr.src (NO_DQS addded)
/sbound.src
/nbonds/nbndcc.src
/nbonds.src
/nbutil.src
/pert/block.src
/icfcnf.src
/rxncor/travel.src
/vibran/quasi.src
/rbquas.src
1.6.4 Parallel Generalized Born Model Code
RMSD-based extension of Rg restraint is added. Parallel gb is fixed
and clean up the code.
Files Modified:
source/energy/energy.src
/genborn.src
/intere.src
/rgy.src
/fcm/rgy.fcm
1.6.5 Cray T3E Related Modifications
LBYCBIM flag is added in calling parameters. NO_BYCU is fixed.
Protected PME_PLAMSA code for non parallel compile. Trajectory
format/unformat is fixed.
Files Modified:
source/nbonds/cenbf.src
/nbndgc.src
/pme.src
/util/util.src
1.6.6 Testcase Fixes
Unit number 99 conflicts with an internal file open unit number,
changed to 89. And other problems causing crash are fixed
Files Modified:
test/c20test/dyntest1.inp
/quasi.inp
/c22test/stdtest.inp
/c25test/hba1.inp
/c26test/pathint.inp
/c27test/mbtest30.inp
------------------------------------------------------------------------------
1.7 Karolinska Institute Fixes and Enhancements
Name : Lennart Nilsson
Email Address : Lennart.Nilsson@biosci.ki.se
Institution : Karolinska Institute
Date : January 31, 2000
1.7.1 Dynamics Restart File Read
New option DYNR [CURR|DELT|VEL] added to READ COOR, allowing
CURRent coordinates, displcement from current (DELT) coordinates or
VELocities to be read from a dynamics restart file.
Files new/modified:
source/io/coorio.src
/dynamc/dynamc.src
doc/io.doc
test/c28test/dynr.inp: NEW
1.7.2 HEAP Expansion
Made HEAP extension working on DEC ALPHA (should also work on other
systems with 64 bit pointers, but right now it is implemented under the pref
directive ALPHA)
File modified:
source/machdep/space.src
1.7.3 OS Information
Modification of OS information in banner to include number of nodes
running the job (from Milan Hodoscek)
File modified:
source/machdep/startup.src
/ctuff.c
1.7.4 Nodenaming Problem Fix
Fixed nodenaming problem on parallel setups using the SOCKET library;
the original version had an inadvertent wild-card capability making nodenames
PE1 and PE10 look identical.
Files modified:
source/machdep/socket.c
1.7.5 COOR COVA/DMAT File Format
Modified output from the COOR COVA and COOR DMAT commands, such that
if results are written to a file, as opposed to the charmm output file,
all headers and blank lines are skipped. This makes it easier to read the
file into a plotting program.
Files modified:
source/manip/corman2.src
1.7.6 Block Data
Fixed block data initialization for the QM interface under GNU
(similar as under HPUX)
Files modified:
source/quantum/qmdata.src
/qmset.src
1.7.7 HBOND Analysis
Fixed error in calcualtion of average number of hbonds printed for
single structure by the COOR HBOND command.
Added HBOND DONOR specifications for residue TIP3 in the test topology
file. This is needed by the hbanal testcases.
File modified:
source/manip/hbanal.src
test/data/top_all22_prot.inp
------------------------------------------------------------------------------
1.8 MBO(N)D Updates
Name : Masa Watanabe
Email Address : watanabe@moldyn.com
Institution : Moldyn, Inc.
Date : December 23, 1999
1.8.1 MOLDYN Include Files
Modifications of MBO(N)D are not correctly implemented in c27b1 and
c28a1. The following include files should be located in two places
source/fcm and source/moldyn. Both places should have exactly the same
files. In c27b1 and c28a1, only source/fcm has the correct version of
these files. Copy source/fcm include files into source/moldyn.
Files involved:
source/moldyn/HANDS2.INC
/HEAPMB.INC
/IFIRST.INC
/INTGRL.INC
/PARAM.FOR
1.8.2 MBO(N)D Fixes and Enhancements
MBO(N)D related bugs have been fixed and BLAS Library dependent code has
been eliminated. MBO(N)D documentation has been updated. c25test/mtsm3.inp
is updated to use atom based cutoff.
Files modified:
build/UNX/Makefile_gnu
/Makefile_sgi
/Makefile_sgi64
/Makefile_ibmrs
source/fcm/mbond.fcm
/mbond/mbdiag.src
/mbmodes.src
/mbond.src
/mbutil.src
/modtrans.src
doc/mbond.doc
test/c25test/mtsm3.inp
The following are introduced and modified during the development period
of February 15, 2000 through August 15, 2000.
2.1 Oxford Physical and Theoretical Chemistry Lab Developments
Name : Ben Webb, and Paul Lyne
Email Address : ben@bellatrix.pcl.ox.ac.uk
Institution : Physical and Theoretical Chemistry Lab, Oxford, UK
Date : July 15, 2000
2.1.1 QM/MM fluctuating charge implementation (FlucQ)
The fluctuating charge model (FlucQ) is implemented in CHARMM and
extended for QM/MM simulation with GAMESS-US, CADPAC or MOPAC. It is designed
primarily for computationally efficient (approx. 10% overhead) modelling of
solvent polarisation in hybrid QM/MM systems. Atomic charges are treated as
extra degrees of freedom, and are propagated by extended Lagrangian methods
during dynamics simulations. Restart and trajectory files are modified when
FlucQ code is active, to store the extra charge information. Extra options have
been added to the COOR ANAL command, to allow the calculation of a dipole
distribution from a trajectory file. The entirety of these changes is activated
by the presence of the FLUCQ keyword in pref.dat.
Files Added:
build/UNX/flucq.mk
source/fcm/flucq.fcm
/flucq/flucq.src, fluqdyn.src, fluqene.src, fluqqmmm.src
Files Modified:
source/cff/enbfast_cff.src, enbonda_cff.src
/charmm/charmm_main.src, iniall.src
/correl/anacor.src, solana.src
/dynamc/dcntrl.src, dynamc.src, dynamc4.src, dynamcv.src, dynio.src
/energy/enefscal.src, energy.src, eutil.src, intere.src, printe.src
/fcm/energy.fcm, param.fcm
/gamint/gamini.src
/image/eimage.src, images.src, upimag.src, xtlfrq.src
/io/parmio.src, trajio.src
/manip/scalar.src
/mbond/mbdiag.src
/mc/mcener.src, mcimge.src
/minmiz/abner.src, conjug.src, egrad1.src, nraph.src, powell.src
/steepd.src, tndriv.src
/misc/hbuild.src
/mmff/enbfast_mm.src, enbscalar_mm.src
/nbonds/cenbf.src, enbfast.src, enbond.src, enbonda.src, enbondg.src
/evdwf.src, ewald.src, grape.src, nbonds.src
/pert/epert.src, icfcnf.src, icpert.src, tsme.src
/quantum/qmene.src
New/Modified Documentation:
doc/corman.doc, flucq.doc[New], scalar.doc
New/Modified Testcase:
test/c28test/fqam1.inp, fqgamess.inp
/data/flucq.crd, flucq.gms, flucq.prm, flucq.rtf
..............................................................................
2.1.2 Changes for compilation with Compaq compilers on Alpha Linux
A compiler switch (FORT) is added to install.com to build CHARMM (and, if
present, CADPAC and GAMESS-US) with the Compaq Fortran and C compilers on Alpha
systems running Linux (use it with "./install gnu medium FORT" or similar).
CHARMM and GAMESS-US are built with 8-byte integers, for proper functioning of
dynamic memory (heap expansion). A new target, "axp-linux", is added to
gmscomp, which requires a version of GAMESS-US recent enough to have
*ALX cards in unport.src (a patch is available on request).
Problems: the assumption that integers are four bytes long is widely spread
throughout the CHARMM code; some of these problems have been fixed here (e.g.
the RXNCOR code) but the main problem is that PSND4 is used to send both
integers and REALs to other nodes, so parallel builds are still broken. In the
long term, perhaps an explicit "pointer" type should be added (via.
preprocessor, probably) and the PSND4 and PSND8 routines replaced with
"send real", "send real*8" and "send integer".
Files Modified:
install.com
build/UNX/Makefile_gnu, gmscomp
source/cadint/cadini.src
/cadpac/intro.F
/fcm/cadpac.fcm
/machdep/space.src
/rxncor/rxndef.src, rxnene.src
..............................................................................
2.1.3 RXNCOR parallel fixes
The RXNCOR code is modified to run certain parts of the code only on
node 0, as it expects to access files (e.g. to print the trace file) which
is generally possible only on this node.
Files Modified:
source/energy/energy.src
rxncor/rxndef.src
..............................................................................
2.1.4 MMFF with g77 fix
A simple one-line fix enables proper compilation of the MMFF code on
systems using the g77 Fortran compiler.
Files Modified:
source/mmff/merckio.src
..............................................................................
2.1.5 CADPAC and QUANTUM fixes, and port to GNU platform
A Makefile is provided for building CADPAC on a GNU machine (Intel or Alpha
Linux). This requires a modified CADPAC distribution (not distributed with
CHARMM) in order to build properly. The install script and Makefile for
compilation on SGI platforms has also been fixed for parallel and/or 64-bit
installations. Although QUANTUM in c28a2 will compile on GNU platforms, it will
not run; a Makefile and modified install script is provided here for its proper
functioning.
Files Added:
build/UNX/cadpac.mk_gnu
/quantum.mk_gnu
Files Modified:
install.com
build/UNX/cadpac.mk_sgi
source/cadint/cadpac/adapquad.F, angmom.f, atoms.F, banner.f, brueckner.F
/chf.f, ddipmpc.f, ddjk.F, dftchf.F, dftmxm2.F
/eigen.F, impt.f, index4.F, intro.F, io.F, jkder.F
/magdft.F, main.F, master.F, mod2.F, mogues.F
/mp2grad.F, mtimit.F, newint.F, nmrs.f, numprop.f
/pack.F, paral.F, quadrature.f, scf.F, sorto.f
/standv.F, start.F, stvder.F, tdhf.F, utilxx.F
/utilyy.f
------------------------------------------------------------------------------
2.2 Enhancements to the Monte Carlo Module and Bugfixes
Name : Aaron R. Dinner
Email Address : aaron@bellatrix.pcl.ox.ac.uk
Institution : University of Oxford
Date : 20 June 2000
2.2.1 MC Enhancements
1) The ACE/ACS solvation energy term is now supported. As described in mc.doc,
exact calculation of the screening energy is highly inefficient, and an
approximate scheme is introduced to accelerate MC simulations which use this
solvation model.
2) It is now possible to minimize the system with the conjugate gradient (CG)
algorithm prior to application of the acceptance criterion. CG is specified
by the keyword "MINI CONJ" in a MOVE ADD call:
MOVE ADD (no mini keywords ) --> no minimization
MOVE ADD (NSTEps > 0 ) --> SD minimization
MOVE ADD (NSTEps > 0, MINI SD ) --> SD minimization
MOVE ADD (NSTEps > 0, MINI CONJ) --> CG minimization
MOVE EDIT should work the same way.
Several parameters for CG minimization are fixed at their default values
(NCGC = 100, PCUT = 0.9999, TOLIter = 100), which could be changed in future
versions if there is interest.
3) The INBFrq keyword is now recognized in the MOVE ADD command and controls
the non-bond list update during minimization. Note that it is separate
from the INBFrq keyword in the MC command.
4) A heuristic non-bond list update has been implemented for MC. The check is
done at the end of each -INBFrq moves.
The frequencies of image list updates (IMGFrq) and energy checks (IECHeck)
need no longer be multiples of INBFrq, but the former will force a non-bond
list update. If either of these parameters is negative, the corresponding
update will be performed whenever a heuristic non-bond list update is done
(and thus should be used only with INBFrq < 0).
5) It is now possible to go through the move set in regular rather than
random order. The keyword to toggle this feature is PICK in MC:
PICK = 0 (default) Random move group (i.e., a MOVE ADD
call) and random move instance within
that group.
PICK = 1 Sequential move group and sequential
move instance within that group.
PICK = 2 Random move group and sequential
move instance within that group.
Note that, for PICK = 2, one does not a priori know how many steps will
cover the move set evenly if the numbers of move instances for different
move groups is different, so there can be a slight end effect. For
example, suppose there are seven steps left and it is time to pick a new
group of move instances; if the move group picked has more than seven
instances, only the first seven will be tried.
Files Added:
source/mc/mcace.src
Files Modified:
source/mc/mc.src, mcener.src, mcmini.src, movead.src, moveio.src
/fcm/mc.fcm
Modified Documentation:
doc/mc.doc
..............................................................................
2.2.2 Bugfixes
1) SKIP NONE
SKIP NONE turns off QM electrostatic terms.
Correct QEQTRM(I) = SKIP to QEQTRM(I) = SKIPN.
Files involved:
source/energy/eutil.src
2) QM/MM VDW energy calculations
QM/MM VDW energies vary with machine due to failure to update group
centers in some cases. A logical variable is passed to an integer argument
(LCENTR) of EGROUP, which results in a machine-dependent cast.
Files involved:
source/quantum/qmene.src
3) Program hangs during first energy calculation
GNU preprocessor options concerning QMMMDATA should be the same as
ALPHA/IRIS/DEC rather than OS2/HPUX. Local variables are not properly
initialized. Add the compiler flag '-finit-local-zero -fno-automatic' to
the FC0 macro.
Files involved:
source/quantum/qmdata.src, quantum/qmset.src
build/UNX/Makefile_gnu
------------------------------------------------------------------------------
2.3 University of Minesota Developments
Name : Cristobal Alhambra, J. Corchado, M. L. Sanchez,
J. Villa, Jiali Gao and D. G. Truhlar
Email Address : alhambra@chem.umn.edu
Institution : U. of Minnesta
Date : June 26, 2000
2.3.1 CHARMM/POLYRATE interface
charmmrate is a module of charmm for interfacing it with
polyrate. The polyrate main program becomes a subprogram of
charmm. polyrate can be called to carry out variational
transition state theory calculations with multidimensional
semiclassical tunneling contributions. When polyrate needs
the value or gradient of the potential energy surface, it
calls a set of interface routines called hooks. The hooks in
turn call charmm routines for energies and gradients
calculated by molecular mechanics or QM/MM methods.
charmmrate-version 1.0 is a module of charmm-version 27
for interfacing it with polyrate-version 8.1.1.
charmmrate-version 1.0 will be distributed as part of the
charmm program, beginning with version 27 of charmm. The
user will also require the crate utility for modifying
polyrate to make it compatible with charmm. The prospective
user of charmmrate should obtain a valid license for charmm
from an authorized charmm licenser and valid licenses for
polyrate and crate from the University of Minnesota
(http://comp.chem.umn.edu).
Files Added:
source/prate/charmmrate.src
/fcm/charmmrate.fcm
build/UNX/prate.mk
Files Modified:
install.com
source/charmm/charmm_main.src
New Documentation:
doc/charmmrate.doc
New Testcase:
test/c28test/cr01.inp, cr02.inp
test/data/cr01p_ini.crd, cr01r_ini.crd, cr01ts_ini.crd,
/cr02.crd, cr02.para, cr02.topo, optcr02.crd
..............................................................................
2.3.2 Semiempirical Valence Bond (SEVB) Term
The SEVB term is a correction to the PES when combined semiempirical
NDDO QM/MM hamiltonians are used to model the reactivity in enzyme active
sites. The code assigns two extended London-Eyring-Polanyi-Sato (LEPS)
potentials to the subsystem {A,B,D} mimicking the acceptor, transferred,
and donor sites, and the necessary parameters entering in the definition of
the Morse functions, i.e. dissociation energies (Dxy), equilibrium distances
(Rxy), beta exponents (bxy), and Sato parameters (Sxy). During a molecular
dynamics simulation the value of the SEVB energy and gradients are evaluated
at every step from the instantaneous configuration of the system and added
to the corresponding arrays.
Files Added:
source/fcm/leps.fcm
/quantum/qmleps.src
Files Modified:
source/quantum/qmset.src, qmene.src
Modified Documentation:
doc/qmmm.doc
New Testcase:
test/c28test/testleps.inp
------------------------------------------------------------------------------
2.4 Biased Molecular Dynamics: Half Quadratic Bias Method (HQBM)
Name : Emanuele Paci
Email Address : paci@pandora.u-strasbg.fr
Institution : Univ. Louis Pasteur (Strasbourg) & Oxford University
Date : Jul 15, 2000
Biased molecular dynamics implementation based on a time dependent half
quadratic perturbation designed to induce large conformational changes on
the simulation timescale and generate low energy pathways.
Files Added:
source/energy/hqbm.src
/fcm/hqbm.fcm
Files Modified:
source/charmm/charmm_main.src
/energy/energy.src
/energy/eutil.src
/fcm/energy.fcm
New Documentation:
doc/hqbm.doc
New Testcase:
test/c28test/hqbm.inp
test/data/1aps.crd, 1aps_mini.crd
------------------------------------------------------------------------------
2.5 Targeted Molecular Dynamics (TMD)
Name : Jianpeng Ma
Email Address : jma@tammy.harvard.edu
Institution : Harvard University
Date : July 20, 2000
This is only for TMD using holonomic RMS constraint. For doing
Targeted Molecular Dynamics (TMD), one needs to define a moving coordinate
and a target coordinate. You slowly pull the moving structure towards
the target structure by gradually decreasing the RMS distance between two.
The 'pulling' speed is defined by user.
Several new keywords for charmm dynamic command (DYNAmic) have been added.
To invoke TMD, very similar to Berendsen's constant temperature method,
one needs to start with the keyword TARGet, then followed by a few others
defined detailed parameters such as the number of atom that subject to
TMD constraint (NTAR), number of dynamic steps for stopping artificial
rotation (INRT), and RMS increment (DINC).
The TMD code can work with and without SHAKE.
Files Added:
source/dynamc/tmd.src
/fcm/tmd.fcm
Files Modified:
source/charmm/charmm_main.src
/dynamc/dcntrl.src, dynamc.src, dynamc4.src, dynio.src
/energy/eutil.src
/minmiz/minmiz.src
Modified Documentation:
doc/dynamc.doc
New Testcase:
test/c28test/tmdtest.inp
------------------------------------------------------------------------------
2.6 BYCC Enhancements
Name : Robert J. Petrella
Email Address : petrella@tammy.harvard.edu
Institution : Harvard University
Date : June 22, 2000
2.6.1 EEF1 Modification
Modified eef1 to be compatible with active atom selections (i.e. with
NBACtive command used in conjunction with BYCC non-bonded option).
Since only active atoms appear in the nonbonded list, the ATOMON array
flags (=1) the atoms appearing in the list. Solvation energies are then
calculated only with the flagged atoms.
Modified Files:
source/misc/eef1.src SUBROUTINE EEF1EN, SLVPRINT and EEF1IM
/energy/energy.src SUBROUTINE ENERGY
/intere.src SUBROUTINE INTER2
doc/eef1.doc
..............................................................................
2.6.2 BYCC code Enhancements
Reduced memory requirements of the BYCC option by assigning arrays
to temporary memory (stack). Eliminated unnecessary common block variables
in actclus.fcm. Also, removed SUBROUTINE RDCLUS from rtfio.src.
Modified Files:
source/nbonds/nbndcc.src SUBROUTINE NBNDGG
/nbonds.src
/io/rtfio.src
/fcm/actclus.fcm
------------------------------------------------------------------------------
2.7 Karolinska Institute Fixes and Enhancements
Name : Lennart Nilsson
Email Address : Lennart.Nilsson@biosci.ki.se
Institution : Karolinska Institute
Date : July 15, 2000
2.7.1 COOR DMAT
A bug in the code for the RMSF option is fixed and the following new
keywords are added:
MATRix to give output in a matrix format, applies to the distance
as well as distance fluctuations
RELAtive to give relative distance fluctuations
DUNIt to give distance matrix output and fluctuations in one go
Files involved:
doc/corman.doc
source/manip/corman2.src
..............................................................................
2.7.2 COOR COVA
The new keyword MATRix is added to give output in matrix format.
Files involved:
doc/corman.doc
source/manip/corman2.src
..............................................................................
2.7.3 COOR HBOND
Modifed output format. With keyword VERBose one now gets listing of
all pairs of hbonds between atoms in the two selections (previously one
would only get statistics for the atoms in the first selection).
Files involved:
doc/corman.doc
source/manip/hbanal.src
..............................................................................
2.7.4 TRAJ READ/WRITe
Documentation is clarified.
Files involved:
doc/dynamc.doc
------------------------------------------------------------------------------
2.8 TSRI Fixes and Enhancements
Name : Charles L. Brooks, III, Michael F. Crowley
Email Address : brooks@scripps.edu, crowley@scripps.edu
Institution : The Scripps Research Institute
Date : July 17, 2000
..............................................................................
2.8.0 General overview. We have introduced a number of improvements
toward making CHARMM functionally more friendly and performance-wise
better, these are described in 2.8.1 - 2.8.5. Additional bugfixes for
particular machines and algorithms are described in 2.8.6 and 2.8.7.
Finally, new developments/improvements for "generalized homology
modeling" (see http://mmtsb.scripps.edu), 2.8.8, and lambda-dynamics
improvements (including enhancements to allow GB to be used with
BLOCK), 2.8.9, are described.
..............................................................................
2.8.1 Fast Shake. This modification involves bring the "fast vector"
SHAKE into operation w/ fast scalar commands to utilize the analytic
3x3 water-specific SHAKE algorithm. This should become the default
for SHAKE in CHARMM, it improves performance when water rigid
molecules are employed very substantially. It is accessed currently
as documented in the SHAKE documentation w/ the shake fast option.
KEYWORD: FSSHK
source/charmm/iniall.src
source/dynamc/dcntrl.src
source/manip/shake.src
..............................................................................
2.8.2 This modification permits access to both a list and substitution
varibles for query regarding the specific pref.dat options used in
building a given CHARMM executable. Documentation appears in
the misc.doc and subst.doc document files.
KEYWORD: PREFMSI
source/charmm/iniall.src
doc/miscom.doc, subst.doc
..............................................................................
2.8.3 This modification was introduced to make permit easier use of
CHARMM on MPI-parallel systems that do not support sdin and stdout.
It turns off stdin and stdout and redirects automatically to files
charmm.inp and charmm.out respectively.
KEYWORD: FILEINPUT
KEYWORD: FILEOUTPUT
source/machdep/machio.src
..............................................................................
2.8.4 Nested Timers to permit more detailed timing information to be
gathered for performance assessment.
KEYWORD: NEWTIMER
source/charmm/charmm_main.src, iniall.src
source/dynamc/dcntrl.src, dynamc.src
source/energy/energy.src
source/image/nbndgcm.src
source/manip/shake.src
..............................................................................
2.8.5 install.com
Added "nolog" command option so log file is not created
and compile output goes to stdout instead - a useful tool for development.
All .mk files changed to enable the "keepf" and "keepo" install.com options.
..............................................................................
2.8.6 Bugfixes for ibmsp machines
source/machdep/paral1.src
..............................................................................
2.8.7 Bugfixes for various problems from c28a2.
KEYWORD: PMEPLSMA
source/mmff/merckio.src
source/nbonds/ace.src, enbfast.src, pme.src, pmeutil.src
source/quantum/qmpac.src
doc/nbonds.doc
..............................................................................
2.8.8 Center of Mass Harmonic Constraints for use in lattice/all atom
refinement with "generalized homology modeling" tools being build by
M. Feig.
KEYWORD: HMCM
source/charmm/iniall.src
source/energy/ecnstr.src, energy.src, eutil.src, printe.src
source/fcm/cnst.fcm, dimens.fcm, energy.fcm
source/manip/cstran.src
source/pert/epert.src
test/c28test/hmcm.inp
test/data/pep5.pdb
doc/cons.doc
..............................................................................
2.8.9 Improvements and new additions to lambda-dynamics code/methods
to permit 1) more general umbrella potentials that are
lambda-dependent be applied for improved convergence and sampling
during lambda-dynamics (See S. Banba et al., JCP, 2000; JPC, 2000); 2)
to allow use of the generalized Born module with the BLOCK and
lambda-dynamics approaches. These are developments from Shinichi
Banba while at TSRI. NOTE: Some of these keywords will change and be
eliminated to more tightly incorporate developments in the next
release.
KEYWORD: BANBA
KEYWORD: LDLAN
KEYWORD: LDMGEN
KEYWORD: LMC
KEYWORD: LRST
KEYWORD: GBBLCK
KEYWORD: GBFIXAT
source/charmm/iniall.src
source/dynamc/dcntrl.src, dynamc.src, dynio.src, dynlng.src
source/energy/eintern.src, enefscal.src, eutil.src, genborn.src
source/fcm/block.fcm, genborn.fcm, lambda.fcm, reawri.fcm, rwlam.fcm
source/image/images.src
source/nbonds/enbfast.src, enbond.src, enbonda.src, nbutil.src
source/pert/block.src
test/c28test/GB_block.inp, GB_fep.inp, GB_ldm.inp, GB_mcmd.inp
/ldm_lang.inp, ldm_lang_slow.inp, ldm_rst.inp, ldm_rst2.inp
/ldm_rst2_slow.inp, ldm_rst_slow.inp
/mcmd_rst.inp, mcmd_rst2.inp, mcmd_rst2_slow.inp, mcmd_rst_slow.inp
/replica_genborn1.inp, replica_genborn2.inp, replica_genborn3.inp
doc/block.doc, dynamc.doc, genborn.doc, pdetail.doc
------------------------------------------------------------------------------
2.9 Slovenia National Institute of Chemistry Developments
Name : Milan Hodoscek
Email Address : milan@kihp6.ki.si
Institution : National Institute of Chemistry, Slovenia
Date : June 23, 2000
2.9.1 Name conflicts resolved
Due to GAMESS compatibility problems,
- NORM subroutine was changed to TQNORM
- EPSLON subroutine was changed to EPSLN
- PSETUP subroutine was changed to PRSET
- PATH subroutine was changed to PATHC
Files modified:
source/charmm/charmm_main.src
/correl/solana.src
/dimb/eispack.src
/dynamc/dcntrl.src, prssre.src
/rxncor/path.src
..............................................................................
2.9.2 GAMESS code update
1) New GAMESS version (includes DFT) replaces gamint/gamess. Parallel
library DDI is partially implemented in CHARMM for better performance.
File Added:
source/gamint/ddi.src
2) New NOGUess option in GAMEss command added. It speeds up some calculations
and make others possible. Default is still the old behavior.
Files modified:
doc/gamess.doc
source/fcm/gamess.fcm
/gamint/gamini.src, blur.src
test/c25test/gmstst.inp
..............................................................................
2.9.3 Header Enhancement
More informational printout in the CHARMM header. It reports the
hostname of the machine (and for parallel also number of additional
processes) and more specific information of the operating system.
Files modified:
source/machdep/cstuff.c, startup.src
..............................................................................
2.9.4 Minor Enhancements
"Y2K" fix and more precision in timings. Format for Volume calcultations
increased.
Files modified:
source/machdep/cstuff.c, machutil.src
/manip/corman2.src
------------------------------------------------------------------------------
2.10 CHARMM / GAMESS-UK Interface
Name : Paul Sherwood
Email Address : p.sherwood@daresbury.ac.uk
Institution : CLRC Daresbury Lab.
Date : July 20, 2000
Ab initio program GAMESS-UK (General Atomic and Molecular Electronic
Structure System, UK version) is connected to CHARMM program in a QM/MM
method. This method is based on the interface to the GAMESS (US version),
the latter being an extension of the QUANTUM code which is described in
J. Comp. Chem., Vol. 11, No. 6, 700-733 (1990).
Some references are renamed in order to resolve conflics between
CHARMM and GAMESS-UK as follows. All other remaining clashes have been
resolved by changing GAMESS-UK source.
common/equiv => common/equiv0 in fcm/cff.fcm
common/deriv => common/derivr in fcm/deriv.fcm
function lstchr => lastch in cff/parrdr_cff.src
function INLIST => QINLIST in image/nbondm.src, nbonds/nbonda.src,
nbonds/nbondg.src, nbonds/nutil.src
ortho1 => ortho1p in rxncor/path.src
shut => chmshut in machdep/socket.c
New Module and Files:
source/gukint
source/gukint/gukini.src
doc/gamess-uk.doc
test/c28test/alanine_guk.inp
/data/alanine_guk.in
Modified Files:
install.com
build/UNX/Makefile_sgi and Makefile_gnu
source/charmm/charmm_main.src, iniall.src
/energy/energy.src
/gener/update.src
/machdep/startup.src
/nbonds/nbonda.src, nbonds.src
/quantum/addlnat.src
Notes:
a) sgi and gnu are the only platforms that the code is tested on.
b) Corrected logical test in mmff/merckio.src.
------------------------------------------------------------------------------
2.11 Combination of soft core potentials and PERT
Name : Stefan Boresch and Michael Brunsteiner
Email Address : stefan@mdy.univie.ac.at
Institution : University of Vienna, Austria
Date : July 5, 2000
The PERT free energy module of CHARMM is based on a linear dependence
on the coupling parameter. While simplifying implementation, this
approach is prone to van der Waals endpoint problems. One widely used
method to overcome the van der Waals endpoint problem is the use of
soft core Lennard Jones and electrostatic interactions for those
energy terms that cause problems. This capability has been added,
following Zacharias, Straatsma and McCammon, J. Chem. Phys. 1994, 100,
9025. A first publication in which a development version of the
code was used is Brunsteiner & Boresch, JCP 2000, 112, 6953-55
Modified Files:
source/charmm/iniall.src initialization
/energy/enefscal.src force use of slow energy routines in subr. FASTST
/nbonds/enbonda.src trap case PSSP; call spec. purpose energy routines
/enbondg.src
/pert/epert.src bulk of changes,
spec. purpose energy routines added
/pert.src parsing, initialization, resetting; adding of add.
dU/dL contributions to standard array entries
/fcm/pert.fcm definitions of nonlocal variables
Modified Documentation:
doc/pert.doc
New/Modified Testcase:
test/c28test/pssp1.inp, pssp2.inp, pssp3.inp
/data/pssp3.pun
------------------------------------------------------------------------------
2.12 Cornell University Developments
Name : Wonpil Im and Benoit Roux
Email Address : Benoit.Roux@med.cornell.edu
Institution : Cornell University
Date : 5 July, 2000
2.12.1 Generalized solvent boundary potential (GSBP)
A generalized solvent boundary potential (GSBP) has been implemented
for simulation of a reduced region while incorporating the influence of
the surrounding implicitly. GSBP is in the same spirit as the SBOUND
potential and the SSBP but is relies heavily on the PBEQ module.
The current implementation of the method is described in W. IM, S. Berneche
and B. Roux. J. Chem. Phys. (2000, to appear).
The GSBP represents the electrostatic forces from the surrounding
as the use of 2 components: One is the static external field which comes
from the charge distribution in the outer region considering the environment
as a featureless dielectric medium. The second one is the reaction field
which is created by the charge distribution inside the inner region
considering the whole molecular configuration and the solvent.
New Files:
source/misc/gsbp.src, gsbp2.src
Modified Files:
install.com (gsbp keyword)
source/energy/energy.src, eutil.src, printe.src
/misc/pbeq.src, pbeq2.src
/fcm/energy.fcm, pbeq.fcm
Modified Documentation:
doc/pbeq.doc
New Testcases:
test/c28test/gsbptest1.inp test the setup of the static field and
reaction field matrix calculation
test/c28test/gsbptest2.inp generate short trajectory with GSBP forces
..............................................................................
2.12.2 Miscelaneous upgrades for the PBEQ module
The PBEQ (Poisson-Boltzmann finite-difference solver) has been upgraded
and improved. Several methods for solving the finite difference equation
have been implemented and tested. Full multi-grid relaxation (FMGR),
optimized successive over relaxation (OSOR), 3D periodic boundary conditions.
Non-linear and partially linearized PB equation have been implemented.
The performance of the code has been tested and improved by comparing with
delPhi. The performance is now fairly similar to that of delPhi.
A complete benchmark input file pbeqtest6.inp is introduced to test all
the convergence methods for a grid of 65 cubic (estimated run time is
about 15 minutes on Pentium III).
A new stream file pka.str has been introduced to calculate pKa of
ionizable residues with all atom PARAM22. The charges of the 2 states
are stored in scalar store 1 and store 2. The usage of the stream file
is illustrated in the pbeqtest7.inp file.
Modified File:
source/misc/pbeq.src, pbeq2.src
/fcm/pbeq.fcm
Modified Documentation:
doc/pbeq.doc
New Testcases:
test/c28test/pbeqtest4.inp test 3D periodic boundary conditions
/pbeqtest5.inp test non-linear and partially linearized PB
/pbeqtest6.inp benchmark of all the finite-difference solvers
/pbeqtest7.inp example of pKa calculation with pka.str
test/data/pka.str, rs12_asp1.crd, str1.crd
..............................................................................
2.12.3 Multiple RMSD (BestFit) harmonic restraints
A new restraint has been introduced to manipulate and control
macromolecular conformations. The new restraint is related to the
CONS HARM BestFit, which sets up harmonic restraints with respect of a
reference structure. However, because all the reference data structure
is stored in XREF, YREF, ZREF, this command allows only a single bestfit
restraint. In addition, it allows only a restraint to zero value of RMSD.
It is useful to allow multiple such bestfit RMSD restraint to progress
from one conformation to a second conformation of a molecular system.
The new command CONS RMSD allows such multiple bestfit restraint.
The form of the new restraint energy is:
E = Sum_i KFORCE_i * [RMSD - OFFSET_i]**2
Where RMSD is the (possibly mass-weighted) root-mean-square-deviation (RMSD)
of the current coordinates with respect to a reference structure,
KFORCE_i is a force constant, and OFFSET_i is a constant value setting
a relative distance with respect to the RMSD of the structure.
The restraint energy is equivalent the normal BestFit energy.
The forces have been checked with the TEST FIRST command.
The syntax is very similar to all current restraints in CHARMM:
CONS RMSD {FORCE } {OFFSET } -
[mass] [comp] [norotation] [notranslation] atom-selection
sets a RMSD new restraint. It should be noted that all the data structure
is stored dynamically on the HEAP and thus, no extra permanent (static)
storage is introduced.
CONS CLEAR
removes all multiple RMSD restraints
CONS RMSD SHOW
prints all current RMSD restraints with all parameters.
New Files:
source/energy/rmsd.src
/fcm/rmsd.fcm
Modified Files:
source/energy/ecnstr.src
/manip/cstran.src
Modified Documentation:
doc/cons.doc
New Testcase:
test/c28test/test_rmsd.inp
..............................................................................
2.12.4 Patch RTF generator for PERT calculations
A new utility command MKPRES has been introduced to write a PATCH
for adding a dummy sidechain onto a backbone with the goal of performing
free energy calculations. The command generates the list of needed dihedral
angles and non-bonded exclusions.
New File:
source/gener/mkpres.src
Modified File:
source/charmm/charmm_main.src
Modified Documentation:
doc/pert.doc
New Testcase:
test/c28test/mkpres.inp
..............................................................................
2.12.5 Bugfix: MMFP GEO PLANAR
It is possible to set a distance restraint between the center of mass
of two atom selection using the GEO restraint. If the keyword planar is used,
the restraint applied only for the projection along the planar normal
(given by xdir ydir zdir , see doc/mmfp.doc).
File Involved:
source/misc/mmfp.src
The following are introduced and modified during the development period
of August 15, 2000 through February 15, 2001.
3.1 Oxford Physical and Theoretical Chemistry Lab Developments
Name : Ben Webb
Email Address : ben@bellatrix.pcl.ox.ac.uk
Institution : Physical and Theoretical Chemistry Lab, Oxford, UK
Date : November 30, 2000
3.1.1 QM/MM fluctuating charge improvements
1) Bugs with freeing memory on shutting down FlucQ, and reading restart files
on parallel machines, are fixed.
2) Charge thermostatting is now supported via. Nose-Hoover, coupling to an
external heat bath, or simple velocity scaling. This is set up with the
FQMASS, FQTCOU and FQSCAL options to the DYNAmics command, respectively.
3) FlucQ simulations are now supported with Ewald (not PME) calculations.
4) An interface to GAMESS-UK is now in place, which requires the standard
GAMESS-UK distribution to be suitably patched.
5) Patches for CADPAC, GAMESS-US, and GAMESS-UK are available in
support/patch to allow FlucQ to function properly with these packages.
Files Modified:
source/flucq/flucq.src, fluqdyn.src, fluqqmmm.src
/charmm/iniall.src
/dynamc/dcntrl.src, dynio.src
/energy/energy.src
/fcm/flucq.fcm
/gukint/gukini.src
/nbonds/enbond.src, ewald.src, ewaldf.src
Files Added:
support/patch/cadpac.patch, gamess-uk.patch, gamess-us.patch
New/Modified Documentation:
doc/flucq.doc
New/Modified Testcase:
test/c28test/fqam1.inp, fqguk.inp
/data/fq_guk.in
..............................................................................
3.1.2 Changes for proper MPI functioning under Alpha Linux (GNUALPHA)
CHARMM must be built with 64-bit integers under Alpha Linux (using the
Compaq C and Fortran compilers, by means of the FORT switch to install.com)
in order for heap expansion and GAMESS to function correctly. However, external
libraries (e.g. for MPI) are generally built with 32-bit integers. Thus, in
order for CHARMM to interface properly with these libraries, all integer
arguments must be explicitly passed as INTEGER*4. Routines that assume
integers to be 4 bytes in length must also be changed to use 8 bytes. The
implementation for MPI under Alpha Linux has been tested for CHARMM linked
with the MPICH libraries (also compiled with Compaq's ccc/fort) on a network
of EV56 machines.
install.com has also been modified to force mpif.h to explicitly define
integers as INTEGER*4 under Alpha Linux. install.com now also uses mpicc/mpif77
rather than gcc/g77 on all GNU systems when MPI is requested.
Files Modified:
install.com
source/machdep/paral1.src, paral2.src, paral3.src
/gamint/ddi.src
..............................................................................
3.1.3 Addition of PM3 parameters to QUANTUM
PM3 parameters are added to the existing AM1 and MNDO parameters for
QUANTUM. They can be used by specifying the 'PM3' option to the QUANTUM
command rather than 'AM1'. Also, by default QUANTUM uses the CHARMM nonbond
cutoffs for the QM/MM interaction (this differs from GAMESS/CADPAC, which
interact with all MM atoms). This behaviour can be disabled, such that
QUANTUM interacts with all MM atoms, by giving the 'NOCUtoff' option to the
QUANTUM command.
Files Modified:
source/fcm/quantm.fcm, am1parm.fcm
/quantum/qmdata.src, qmene.src, qmpac.src, qmset.src, qmnbnd.src
New/Modified Documentation:
doc/qmmm.doc
New/Modified Testcase:
test/c28test/pm3.inp
..............................................................................
3.1.4 Addition of a "simple combination" to RXNCOR code
The existing RXNCOR "COMBination" command is supplemented by a "simple"
combination ("SCOMbination") command. This takes a simple weighted sum
of all names, as opposed to a weighted mean, and thus allows negative weights
for tracing (for example) the difference between two bond lengths.
Files Modified:
source/rxncor/rxndef.src
New/Modified Documentation:
doc/umbrel.doc
------------------------------------------------------------------------------
3.2 University of Minesota Developments
Name : Jiali Gao
Email Address : gao@chem.umn.edu
Institution : University of Minnesota
Date : December 29, 2000
3.2.1 Semiempirical Quantum-Molecular Mechanical Code Enhancements
1) QM/MM electrostatic free energy perturbation
A new option is added to the QUANtum command to perform
free energy calculations for QM/MM interactions.
The PERT command performs electrostatic free energy decoupling
calculations for QM/MM interactions on the fly of a molecular dynamics
simulation. The algorithm is based on a method described in J. Gao,
J. Phys. Chem. 96, 537 (1992). Through a series of simulations, the
electrostatic component of the free energy of solvation can be determined.
2) QM/MM interaction energy decomposition
The DECOmposition option is added to the QUANtum commad, which
initiates an qm/mm interaction energy decomposition calculation
on the fly during a molecular dynamics simulation using the QUANtum command.
It is currently implemented only for semiempirical Hamiltonians.
The analysis is based on the method reported in J. Gao and X. Xia,
Science, 258, 631 (1992). It decomposes the total QM/MM electrostatic
interaction energy into a vertical interaction energy Evert, and a
polarization term Epol. The latter is further separated into
electrostatic stabilization Estab, and charge distortion Edist.
3) Modification of space allocation for qm/mm interactions
Reduce the space allocated for qm/mm interactions. Previously, space
for all mm atoms are allocated. Now, it is reduced to only request space
for atoms that actual interact with qm atoms. This only affects the
semiempirical code.
Files Modified:
source/quantum/qmene.src, qmnbnd.src, qmset.src
/dynamc/dynamc.src
/energy/eutil.src
Modified Documentation:
doc/quantum.doc
New Testcase:
test/c28test/qmlep.inp
Note that qmfep.inp is basically the same input for the qmmm testcase
c22test/quantum1.inp
..............................................................................
3.2.2 Umbrella Sampling Free Energy Calculation
A biasing potential for umbrella sampling free energy
calculations (form 5) is added.
RXNCord: UMBRella FORM form KUMB ku DEL0 del0
form functional form of potential
---- ----------------------------
1 ku*(delta-del0)**2
5 ku*(delta-del0)**2 + Ubias(delta)
Ubias is a biasing potential, used in addition to the harmonic restraining
potential, to truncate high barriers of activated processes. Ideally,
Ubias(Rc) would be the negative of the potential of mean force g(Rc),
where Rc is the reaction coordinate. Ubias is implemented as a cubic spline
function based on a tabulated data to be read prior to the call of RXNC.
Files Modified:
source/rxncor/rxndef.src, rxnene.src
Modified Documentation:
doc/umbrel.doc
------------------------------------------------------------------------------
3.3 Enhancements to the Monte Carlo Module
Name : Aaron R. Dinner
Email Address : aaron@bellatrix.pcl.ox.ac.uk
Institution : University of Oxford
Date : 30 November 2000
1) Hybrid Monte Carlo (HMC) moves were added. In such moves,
the positions of the atoms are propagated with a molecular
dynamics integrator (DYNAMC). See mc.doc and references
therein for further description.
2) Rigid body translations and rotations of heavy atoms and
their associated hydrogen atoms (RTRN/RROT BYHEavy) were
added. Such moves allow one to compare MC simulations with
dynamics simulations constrained with SHAKE BONH directly.
For consistency, the CART key word was changed to RTRN BYATom;
the subroutines MVRTRN and MVCART were combined to condense the
code.
3) Group-based energy calculations now scale correctly with
system size. This modification involved changing the calling
arguments of EGROUP, ENBFSG, ENBFVG, and EIMNBD, which affected
several source files (detailed below).
4) Concerted rotations that select the original conformation
(to maintain detailed balance) are now counted as rejections
for ARM, DOMC, and MCSTAT. Also, the correct original driver
angle rotation is now passed back. Both these changes should
improve automatic optimization of CROT moves.
5) The calling arguments of EEF1EN were changed to accelerate MC
calculations using this solvation term and for consistency with
the group-based energy calculation changes described above. Also,
the call to EEF1EN in MCENER was fixed to include the ATOMON list.
6) The LCENTR flag is now handled correctly in ENBFVG.
Files Added:
mc/mvhmc.src (1)
Files Modified:
source/dynamc/dynamc.src (1)
/energy/energy.src (3,5), intere.src (5)
/image/eimage.src (3)
/mc/mc.src (1,2,3,4), mcace.src (3), mcener.src (3,5)
/mcimge.src (3), movead.src (1,2), moveio.src (1)
/mvcrot.src (4), mvrtrn.src (2)
/misc/eef1.src (5)
/nbonds/enbfast.src (3,6), enbond.src (3), enbondg.src (3)
/pert/epert.src (3)
/fcm/reawri.fcm (1)
Modified Documentation:
doc/mc.doc
------------------------------------------------------------------------------
3.4 TSRI Fixes and Enhancements
Name : Charles L. Brooks, III and Michael F. Crowley
Email Address : brooks@scripps.edu, crowley@scripps.edu
Institution : The Scripps Research Institute
Date : December 29, 2000
3.4.1 Installation Enhancements
gukint is added to make. ibmsp3 Makefile is added. NOLOG argument
is added to send compile output to stdout rather than a log file.
Files Added/Modified:
install.com
build/UNX/Makefile_ibmsp3 [New]
..............................................................................
3.4.2 Grid based energy calculation
A new function is added to calculate grid-based energies.
Files Modified:
source/charmm/charmm_main.src, iniall.src
/energy/energy.src, eutil.src, intere.src, printe.src
/fcm/energy.fcm, grid.fcm [New], pbeq.fcm
/misc/grid.src [New], pbeq.src, pbeq2.src
New Documentation:
doc/grid.doc
..............................................................................
3.4.3 Timing Calls
The new timer function is extended and runs on parallel machines.
Files Modified;
source/dynamc/dcntrl.src, dynamc.src, dynamc4.src, dynamvv.src
/energy/energy.src
/util/new_timer.src
/fcm/new_time.fcm, parallel.fcm
/image/nbndgcm.src
/machdep/machio.src, paral1.src, paral2.src, paral3.src
/nbonds/enbond.src, ewaldf.src, heurist.src
..............................................................................
3.4.4 Fast SHAKE
Fast shake (analytical) for all 2, 3, and 4 body constraints.
Set up for all bonh (TIP3, ch, ch2, ch3, nh, etc), Very much improved
fast shake for non-vector machines. Works for parallel. (keyword FSSHK)
Files Modified:
source/fcm/fstshk.fcm
/gener/update.src
/manip/fsshake.src, shake.src
..............................................................................
3.4.5 Bugfixes and Enhancements
doc/cons.doc - insignificant change concerning fast shake
/galgor.doc - update for bugfix
source/fcm/cff.fcm - ##IF CFF added to save huge amounts of memory
when CFF keyword is not used
/fcm/dimens.fcm - increased maximum for number of dihedrals for LARGE
/fcm/galgor.fcm - changes for ic seed atom specifications
/gukint/ukini.src - added ##IF GAMESSUK
/io/parmio.src - added ability to have geometric combining rules for
vdw parameters enabling OPLS FF to be implemented
in charmm (keyword: GEOMVDW)
/misc/genetic.src - changes for ic seed atom specifications
/mmff/enbfast_mm.src - flucq error fixed: removed offending endif
/nbonds/nbexcl.src - makitc_clr() routine added to clear image memory
/nbonds/nbonds.src - NO_BYCC logic added
/nbonds/nbutil.src - bugfix for softvdw
/pert/block.src - fixed allocation of ibvcrrt
/util/prpref.src - added FLUCQ
..............................................................................
3.4.5 Testing
Lots of test cases changed to fit the formula of being able to stream,
get rid of references to unit 5 and unit 99. Check keywords to decide
if the test case is appropriate for the pref.dat keywords that were used
for the executable being tested. Made run parallel jobs on mpi, t3e, ibmsp3,
and ibmsp.
Files New/Modified:
test/test.com
/stream.inp [New]
------------------------------------------------------------------------------
3.5 Slovenia National Institute of Chemistry Developments
Name : Milan Hodoscek
Email Address : milan@kihp6.ki.si
Institution : National Institute of Chemistry, Slovenia
Date : December 28, 2000
3.5.1 EEF1 method is parallelized
Files Modified:
source/misc/eef1.src
Modified Documentation:
doc/parallel.doc
..............................................................................
3.5.2 Arrays references
Some arrays in common are used in too large loops, and some access the
zeroth element, etc....
Files Modified:
source/energy/eutil.src - line 388: should be DO I=1,LENENP
/io/rtfio.src - line 263: rtrtyp(nrtrs); but nrtrs is put to 0
/util/matrix.src - lines 90,245,285,377,822,1018,1058:
A(1) -> A(*) changes
/util/title.src - lines 57,59: limit problems in LINE()
..............................................................................
3.5.3 Energy printout with QM/MM
QM energies are converted from Hartrees to kcal/mole and become very
large numbers which energy/printe.src couldn't handle. Fixed!
File Modified:
source/energy/printe.src
..............................................................................
3.5.4 Common block names
Some of the names in .fcm SAVE statements and COMMON statements don't
agree. All such occurences are fixed.
Files Modified:
source/fcm/fft.fcm
/fcm/intcr.fcm
/fcm/mmff.fcm
..............................................................................
3.5.5 XXLARGE size keyword
install.com xxlarge didn't work. Fixed.
Files Modified:
install.com
source/fcm/heap.fcm
/fcm/stack.fcm
..............................................................................
3.5.6 Performance problem
Some of the common blocks are not aligned for efficient performance.
The speed of CHARMM runs varies if compiled with different pref.dat keywords.
File Mofdified:
source/fcm/lambda.fcm
..............................................................................
3.5.7 Miscellaneous small bug fixes
Files Modifed:
source/fcm/lobos.fcm - wrong position of ##ENDIF
/gamint/ddi.src - typo in comment statement
/gamint/gamini.src - correct message for blur method
/image/nbndgcm.src - line 644: xw(3,nat) must go away!
/machdep/startup.src - line 229: allow use of parameters for any
parallel
/manip/shake.src - line 271: call parptupdate() should be
removed!
/mmff/enbfast_mm.src - line 81: This ENDIF must be deleted!
/nbonds/pme.src - line 357: latm problem in parallel
install.com - changed: reduced -> reduce as it is in
dimens.fcm
Modified Documentation:
doc/gamess.doc - minor corrections
------------------------------------------------------------------------------
3.6 Encapsulation of the PERT/PSSP (Soft Core) Code
Name : Stefan Boresch
Email Address : stefan@mdy.univie.ac.at
Institution : University of Vienna, Austria
Date : December 17, 2000
The PSSP code (combination of PERT with soft core potentials) added in
c28a3 has been put in ##IF PSSP brackets so that it can be cleanly
eliminated if not needed. Removing PERT from pref.dat has the same
effect (PSSP without PERT is useless). No functional changes were
introduced! PSSP is needed in the pref.dat to activate the PERT/PSSP
code.
Files Modified:
source/charmm/iniall.src
/energy/enefscal.src
/fcm/pert.fcm
/pert/epert.src, pert.src
------------------------------------------------------------------------------
3.7 New Random Number Generator
Name : Xiaoliang Qian, Dan Strahs and Tamar Schlick
Email Address : qian@biomath.nyu.edu
Institution : New York University
Date : December 22, 2000
A new Random number generator (RNG) is provided based on
Pierre L'Ecuyer & Terry H Andres' C version code (see clcg.src).
This RNG combines 4 linear congruential generators (LCGs)
to get the long period of about 2^121 for the resulting sequence.
Multiple independent streams can also be achieved for parallel
computing. The current state for each stream is stored by four
integers rather than one for the old RNG.
A representative usage of the CLCG RNG and the replacement
of the old RNG is needed for MD with the LN integrator, which
makes larger systems and longer simulation durations accessible,
exhausting the period of the old RNG.
To conform to the existing function calling interface, the new RNG
accepts a variable ISEED. However, the new RNG actually requires
the propagation of four previous states, not just one state; hence,
ISEED is now interpreted to select one of several random number streams,
ranging from 1 to . (Maxgen is defined in fcm/clcg.fcm).
Since ISEED is used to access a particular stream, the default value of
ISEED (in iniall.src) will be 1 rather than 314159. Call from subroutines
attempting to specify ISEED are reinterpreted as modulo(Maxgen) to select
a the random number stream. Note that a separate subroutine SetSeed
(clcg.src) is included for setting the initial seed of the new RNGs.
The new RND code is activated with the pref key "NEWRNG". The default
is still the "old" RNG. Old dynamics restart files are not compatible
with the new RNG code. (see dynamc/dynio.src)
Multiple streams requires the propagation of 4*Nstream seeds through
restart files, where Nstream is the number of streams. This suggests that
the fixed format of the restart file would become variable. A possible
solution emerges with adding an extra line or variable indicating the
number of streams and, hence, the number of seed lines expected to be read.
Files Added:
source/util/clcg.src
/fcm/clcg.fcm
Files Modified:
source/charmm/iniall.src
/util/random.src
/dynamc/dynio.src
The following are introduced and modified during the development period
of February 15, 2001 through July 15, 2001. c28b0 is the pre-release version
tested prior the c28b1 release on August 15, 2001.
4.1 TSRI Fixes and Enhancements
Name : Charles L. Brooks, III and Michael F. Crowley
Email Address : brooks@scripps.edu, crowley@scripps.edu
Institution : The Scripps Research Institute
Date : July 2, 2001
4.1.1 install.com
Pref keys GBMF, GRID and RCFFT are added. Note that RCFFT works
only with PARALLEL. SCALI MPI installation is added and SP3 FSSCK
keyword error is fixed.
4.1.2 Genborn
Added new GB module to allow individual atom-type based
parameterization of Still-like GB model. Still under development.
Files modified:
doc/genborn.doc
source/charmm/charmm_main.src
/energy/energy.src, genborn.src, intere.src
/fcm/genborn.fcm
/machdep/cstuff.c
4.1.3 No Buffering
Added keyword to eliminate output buffering.
Files modified:
source/machdep/cstuff.c
4.1.4 Grid-Based Potentials
Additional development of grid-based potentials for ligand-receptor
docking studies. Modifications and testcases added.
Files modified:
source/charmm/iniall.src
/energy/energy.src, intere.src
/misc/genetic.src, grid.src
/util/prpref.src
New testcases:
test/c28test/gridtest.inp, grid_2.inp
/data/3ptb_complex19.crd, 3ptb_complex19.psf, grid_3ptb.bin
4.1.5 Real-to-Complex FFT for PME
Using real_to_complex FFT method uses the symmetry of the transformed
charge matrix to cut memory requirements in about 1/2 and computation down
by about 1/3 or more. Only parallel implementation at present. Speedup of
50% on the FFTs results in 5% overall speedup for dhfr benchmark.
pref.dat keyword: RCFFT
Files modified:
source/nbonds/nbutil.src, pme.src, pmeutil.src
4.1.6 Faster Direct Sum
Rearranged loops, moving if statements outside loops (amounts to the
same thing as expand) and prescreening the atoms inside the cutoff before
doing the 6/12 and erfc loop.
pref.dat keyword: FASTEW
Files modified:
source/nbonds/ewald.src
4.1.7 IEEE 64 Bit Trajectory Reading
This is to read t3e/j90 trajectories from SGIs.
pref.dat keyword: T3ETRAJ
Files modified:
source/dynamc/dynio.src
/fcm/stream.fcm
/util/title.src
4.1.8 Compilation
Compilation feature for F90 and N32 added for SGIs
pref.dat keyword: SGIF90
Files modified:
install.com
source/machdep/machio.src, machutil.src, paral2.src, paral3.src
/mmff/datastruc.src
4.1.9 test.com
Testcases made run on parallel platforms. Need to change file names
into lower case only. Also added checking features compiled.
Files modified:
test/test.com
/c20test/brbtest.inp, dyntest2.inp, maatest.inp, trnphi.inp,
/vibpafl.inp, vibran.inp, vibwat.inp, vibrtst.inp
/c22test/cortst.inp, icfix.inp, icpert.inp, mewtest.inp,
/path.inp, pert.inp, slowgr.inp, stdtest.inp, window.inp
/c23test/clustst.inp
/c24test/pert2.inp
/c25test/cortst25.inp, helix.inp, pull.inp, quiet.inp, resdtest.inp,
/vibwat25.inp
/c27test/cubesgb.inp, genborn19.inp
/c28test/alanine_guk.inp, cr01.inp, cr02.inp
Files renamed:
test/c24test/dyn4dtest.inp
/c27test/gb_cff-data.inp, gb_cff95_na-test.inp, gb_cff95_prot-test.inp,
/gb_mmff94_na-test.inp, gb_mmff94_prot-test.inp,
/gb_mmff94_test.inp
/c28test/gb_block.inp, gb_fep.inp, gb_ldm.inp, gb_mcmd.inp
------------------------------------------------------------------------------
4.2 University of Minesota Developments
Name : Jiali Gao
Email Address : gao@chem.umn.edu
Institution : University of Minnesota
Date : June 21, 2001
(1) allowing qm/mm group-group list for mm-group and qm-molecule
interactions in qm/mm calculations.
(2) efficient management of memory allocation for qm/mm.
(3) Inclusion of qm/mm periodic bondary conditions for qm/mm
calculations using Charlie Brooks' PBound.
Files Modified:
source/fcm/nbndqm.fcm
/quantum/qmene.src, qmnbnd.src
------------------------------------------------------------------------------
4.3 CFF Enhancements
Name : Nan-Jie Deng
Email Address : ndeng@msi.com
Institution : Molecular Simulations Inc.
Date : April 20, 2001
Second derivatives related bugs/problems in CFF have been fixed and
CFF documentation has been updated. This bugfix involves fixing the
calculation of analytical second derivatives in the CFF force field.
As a result, methods that uses analytical second derivatives such as
Newton-Raphson minimization and normal modes calculation are now supported
with the use of CFF force field.
Files modified:
source/fcm/cff.fcm
/cff/codes_cff.src, efscalar_cff.src, enbfast_cff.src,
/enbonda_cff.src, escalar_cff.src, parrdr_cff.src
/charmm/iniall.src
/util/parse.src
/io/rtfio.src
doc/cff.doc
------------------------------------------------------------------------------
4.4 Harvard University Developments
Name : Robert J. Petrella
Email Address : petrella@tammy.harvard.edu
Institution : Harvard University
Date : June 21, 2001
4.4.1 ESTATS
Command "ESTATS"-- utility for calculating average and standard
deviation of the potential energy and its components "on-the-fly"--
i.e. during the actual simulation or sampling run, without the need
for post-processing of trajectory files.
pref.dat keyword: ESTATS
Files modified:
doc/energy.doc
source/charmm/charmm_main.src, iniall.src
/energy/anal.src, energy.src
Files added:
source/fcm/estats.fcm
4.4.2 ADUMB Modifications
"CORR" keyword--facility for "on-the-fly" calculation of the
average values of specified variables as a function of the reaction
coordinates in adaptive umbrella sampling runs.
Files modified:
doc/adumb.doc
source/adumb/cmds.src
/umb.src
/charmm/iniall.src
Files added:
source/fcm/umbcor.fcm
4.4.3 TSM modifications
"RUNA" keyword, used in conjunction with "SAVIc" keyword--
facility for "on-the-fly" calculation of the free energy and average
energy values during internal coordinate perturbation.
Files modified:
doc/perturb.doc
source/charmm/iniall.src
/fcm/icpert.fcm
/pert/tsms.src, icpert.src
------------------------------------------------------------------------------
4.5 Graphics Module Enhancement
Name : Rick Venable
Email Address : Rick_Venable@nih.gov
Institution : FDA/CBER Biophysics Lab
Date : July 16, 2001
4.5.1 Added TRAJ commands to GRAPHX
Historically, only "COOR READ ... IFILE n" could be used within GRAPHX
to read trajectory frames. Support for TRAJ READ greatly facilitates
viewing trajectories or making molecular animations at other than the
default SKIP value from the trajectory file.
Files Modified:
source/graphics/graphx.src
doc/graphx.doc
------------------------------------------------------------------------------
4.6 Bug Fixes and Enhancements
Name : Milan Hodoscek
Email Address : milan@cmm.ki.si, milan@helix.nih.gov
Institution : National Institute of Chemistry, Slovenia
Date : July 16, 2001
4.6.1 Added support for MD-GRAPE-II board
The new version of GRAPE, which is 10 times faster then previuos
versionis now supported.
Files Modified:
source/charmm/iniall.src
/nbonds/enbond.src, grape.src, nbutils.src
/fcm/grape.fcm
4.6.2 Compiler fixes
- Complaints about the standard routine names SETUP, MERGE
- preflx keyword missing
Files Modified:
source/mmff/mmff.src (SETUP -> SETUPMF)
/dynamc/dynsub.src (MERGE -> TMERGE)
/charmm/charmm_main.src (MERGE -> TMERGE)
/fcm/noe.fcm (!##PNOE missing)
4.6.3 Bug Fixes
RGYR command didn't pass TEST FIRSt
MINI CONJ fails to converge - bug fixed
Files Modified:
source/energy/rgy.src (RGYR fix)
/minmiz/conjug.src (MINI CONJ fix)
4.6.4 Documentation updated
Files Modified:
doc/parallel.doc, gamess.doc, developer.doc
