ChangeLog.c35
|
| | |
|
About CHARMM35 Development |
|
|
|
At the July 2007 CHARMM meeting (University of Maryland, Baltimore County
July 20-22, 2007), we decided to release CHARMM c34b1 with incorporating
enhancements and fixes to the developmental version c34a2. The modified
source code and features are described in ChangeLog.c34, [3] c34b1 Change Log.
As an on-going project, CHARMM development will be carried out
with CHARMM version 35a series. The following describes new features
and enhancements incorporated in the c35 development period.
The following are introduced and modified during the development period
of February 15, 2007 through August 15, 2007. The developments are
based on CHARMM version c34a1. Some enhancements and fixes to c34a2
are also incorporated into c34b1 as described in the c34b1 release note
(ChangeLog.c34).
1.1 TSRI Developments
Developers : Mike Crowley, Jana Khandogin, Ilja Khavrutskii,
Charles L. Brooks, III
Email Address : michael_crowley@nrel.gov, janakhan@scripps.edu,
ikhavru@mccammon.ucsd.ed, brooks@scripps.edu
Institution : The Scripps Research Institute
Date : July 5, 2007
1.1.1 Column fft for better parallel performance in PME
Better code for parallel column fft to reduce communications for high
processor counts and increase load balancing.
Prefx Keys: COLFFT
Files Modified:
source/nbonds/nbutil.src
source/nbonds/pme.src
source/nbonds/pme_column_fft.src
source/nbonds/pme_column_recip.src
source/nbonds/pmeutil.src
test/c26test/pme_p21.inp
..............................................................................
1.1.2 Image facility added to the PHMD module
Constant pH molecular dynamics can be now performed for periodic boundary
systems.
Files Added:
test/c35test/phmd_images.inp
test/data/phmd_hsp.in
Files Modified:
doc/phmd.doc
source/energy/energy.src
source/energy/phmd.src
..............................................................................
1.1.3 Harmonic Fourier Beads Method
Harmonic fourier bead method for transition path optimization and
identification.
Files Added:
doc/hfb.doc -->> missing
test/c35test/trek_hfb.inp
Files Modified:
source/rxncor/adiab.src
source/rxncor/travel.src
source/rxncor/travel2.src
..............................................................................
1.1.4 RMSD constraint code changed to module
RMSD code changed to module to aid in revitalizing the second-selection
mode.
Files Modified:
source/dynamc/dcntrl.src
source/energy/ecnstr.src
source/energy/rmsd.src
source/manip/cstran.src
..............................................................................
1.1.5 Various fixes and enhancements
Files Modified:
source/correl/rdfsol.src uninitialized variable fixed
source/dynamc/dynamc.src io control fixed
source/dynamc/dynamvv.src io control fixed
source/energy/gbsw.src deallocation of unallocated array
source/manip/shake.src typo in pref keyword MPI_AIX64 fixed
source/nbonds/enbond.src getmaxnprs() fix for Intel compiler
source/nbonds/ewald.src getmaxnprs() fix for Intel compiler
source/nbonds/pme_column_recip.src cheq module migration error fixed
source/util/parse.src Added ?iseed as query variable
------------------------------------------------------------------------------
1.2 Re-introduction of GBMV Additions into Fortran 90 Module
Developer: : Michael Feig
Email Address : feig@msu.edu
Institution : Michigan State University
Date : June 30, 2007
During the switch of GBMV to Fortran 90, recent additions to the
GBMV module were lost (in particular the VDWaals code). This update
re-introduces previously submitted updates.
Files Modified:
doc/gbmv.doc
source/energy/gbmvmodule.src
------------------------------------------------------------------------------
1.3 University of Chicago Developments
Developers : Edward Harder, Yuqing Deng and Benoit Roux
Email Address : eharder@uchicago.edu, deng@mcs.anl.gov, roux@uchicago.edu
Institution : University of Chicago, Argonne National Lab
Date : July 3, 2007
1.3.1 New Format for Topology/Parameters of Drude Mode
A new format encodes the Drude model from the RTF. In the present
implementation the drude particles are generated for all atoms for which
polarizabilities (ie. via ALPHA) are specified in the RTF file.
This allows for the drudes to be generated automatically when a molecule
is generated in CHARMM. In addition, code has been developed to allow for
inclusion of atom-based Thole scale factors, atom-based anisotropic
polarizabilities and the addition of lone pairs to selected atoms at
the RTF level. These enhancements allow for all the information for the
polarizable drude force field to be included in the RTF and parameter files.
Files Removed:
source/fcm/thole.fcm
Files Modified:
doc/drude.doc
doc/fitcharge.doc
doc/tpcntrl.doc
doc/lonepair.doc
source/charmm/charmm_main.src
source/charmm/iniall.src
source/energy/energy.src
source/fcm/aniso.fcm
source/fcm/dimens.fcm
source/fcm/psf.fcm
source/fcm/rtf.fcm
source/gener/drude.src
source/gener/genpsf.src
source/gener/modpsf.src
source/gener/update.src
source/io/parmio.src
source/io/psfres.src
source/io/rtfio.src
source/manip/corman.src
source/misc/fitcharge.src
source/nbonds/nbexcl.src
source/pert/epert.src
source/util/array.src
..............................................................................
1.3.2 Various updates and enhancements
New GSBP cavity potential for the sphere calculation is introduced.
A WCA repulsive potential is placed before the dialectric boundary to prevent
water molecules going out of the atomistic simulation region.
In the generalized multiple calculation, all the charges are included (even if
they could be outside the atomistic region) to avoid discontinuity in the
electrostatic potential.
Read and write the multipole expansion coefficients and external field
(PHIX) in card format. Fixed a bug when the large box calculation has more
grid points than the focused box in GSBP setup calculation.
Exclude GCMC ghost atoms from the extended electrostatics calculation and
from the single distance restraint potential. Removed the temporary WCAD array
on the stack and passed WCAD from psf in QCGSET and CGINV. Increase the number
of maximum selection definitions to 200 in fcm/selcta.fcm
Files Modified:
doc/pbeq.doc
source/energy/energy.src
source/fcm/selcta.fcm
source/misc/pbeq2.src
source/misc/gsbp.src
source/misc/mmfp.src
source/nbonds/nbondg.src
source/nbonds/exelec.src
source/nbonds/cheqmodule.src
------------------------------------------------------------------------------
1.4 QUB (QM/MM path integral) Updates
Developers : Dan Thomas Major and Jiali Gao
Email Address : majort@mail.biu.ac.il and gao@chem.umn.edu
Institution : Bar-Ilan University, Israel and University of Minnesota
Date : June 30, 2007
1.4.1 Enhancement to QUB module
The code has been significantly updated and now includes three submodules
(QCP,BQCP, and QFEP). QCP performs standard Quantized Classical Path, BQCP
performs QCP with bisection sampling, while QFEP performs mass-perturbation
BQCP. The latter allows for accurate estimates of kinetic isotope effects in
condensed phase reactions (See doc file for some tech. references).
1.4.2 Path-Integral coupling with QM/MM methods
The code is coupled with the QUANTUM, SQUANTM, and SCCDFTB modules.
In particular, the coupling with QUANTUM and SQUANTM is tight, and a number
of efficient enhancements have been implemented.
With QUANTUM and SQUANTM gradient calculations are automatically turned off
for efficiency, and for QUANTUM the 1- and 2-electron integrals for QM atoms
not described as path-integrals are stored by QUB to avoid recalculation
during path-integral sampling. This will be done for SQUANTM in the near
future as well.
1.4.3 Parallel Code
QUB runs in parallel with both QUANTUM and SQUANTM. QUB takes control
of parallel run instead of QUANTUM and SQUANTM, as it scales nearly linearly
with number of processors. QUB doesn't run in parallel with SCCDFTB, yet.
Files Added:
test/cquantumtest/qmpathint1.inp
test/cquantumtest/qmpathint2.inp
test/cquantumtest/sqmpathint1.inp
test/cquantumtest/sqmpathint2.inp
Files Modified:
doc/parallel.doc
doc/qub.doc
source/charmm/charmm_main.src
source/charmm/iniall.src
source/fcm/qubpi.fcm
source/fcm/squantm.fcm
source/quantum/qub.src
source/quantum/mpiqub.src
------------------------------------------------------------------------------
1.5 Plastic Network Model (PNM) Implementation
Developmers : Jingzhi Pu, Paul Maragakis, and Martin Karplus
Email Address : pu@tammy.harvard.edu; Paul.Maragakis@deshaw.com;
marci@tammy.harvard.edu
Institution : Harvard University
Date : July 1, 2007
The plastic network model (PNM) gives a simple description of the free
energy landscape that has multiple basins around meta-stable conformers.
The PNM energy can be viewed as an EVB type potential based on two diabatic
states expressed as elastic networks of the stable conformational states.
The coupling of the two states is described by a mixing constant. The method
has been applied to adenylate kinase to generate transition pathways for the
conformational change between the open and closed states. See references
given below for details:
[1]. Maragakis, P.; Karplus, M. J. Mol. Biol. 2005, 352, 807.
[2]. Tirion, M. M. Phys. Rev. Lett. 1996, 77, 1905.
The current implementaion of PNM in CHARMM is limited to case with two
end states, where the user can specify different spring constants, network
cutoffs, and zeroes of energy separately for the two states (see doc/pnm.doc).
The current implementation also allows the user to define two subgroups of
atoms, where the inter-network connection between the two subgroups can be
removed, and only the intra-network connections are retained. This function
may be useful in studying the coupling between different subunits in a
multi-subunit protein complex.
Files Added:
doc/pnm.doc
source/fcm/pnm.fcm
source/misc/pnm.src
test/c35test/test_pnm.inp
test/data/bmfe.top
test/data/bmfe.par
test/data/bmfe.psf
test/data/bmfe.crd
test/data/bmff.crd
Files Modified:
source/charmm/charmm_main.src
source/energy/ediff.src
source/energy/energy.src
source/energy/eutil.src
source/energy/printe.src
source/fcm/energy.fcm
------------------------------------------------------------------------------
1.6 GHO Method Interfac to SCC-DFTB (GHO-SCCDFTB)
Developers : Jingzhi Pu and Jiali Gao
Email Address : pu@tammy.harvard.edu; gao@chem.umn.edu
Institution : Harvard University
Date : July 1, 2007
The origianl implementation of GHO-SCCDFTB in c34a2 has a restriction
that all GHO boundary atoms have to be placed at the end of the QM
fragment. This limitation is inherited from an earlier GHO implementation
for the semiempirical module QUANTUM in CHARMM (see doc/qmmm.doc,
GLNK: Restrictions; a common practice is to write a new residue for
QM atoms and place GHO boundary atoms at the end of that residue).
This restriction is now removed by rearranging the list of QM atoms
internally in CHARMM. The source files need to be modified are:
A bug has been fixed for GHO-SCCDFTB gradient. This bug caused
wrong gradient if the compiler does not initialize a real type
variable to zero when it is declared (for example, g77 under Linux).
We fixed this bug by initializing some variables explicitly.
Files Added:
test/cquantumtest/buta_ghosccdftb.inp
Files Modified:
source/sccdftbint/sccdftbini.src
source/sccdftbint/stbgho.src
------------------------------------------------------------------------------
1.7 Semiempirical QM/MM SQUANTM and MNDO97 Updates
Developers : Kwangho Nam
Email Address : nam@tammy.harvard.edu
Institution : Harvard University
Date : June 30, 2007
1.7.1 SQUANTM Fortran90 format change
The format of SQUANTM semiempirical QM/MM module is changed to
comply with other CHARMM source format, allowing other CHARMM fcm
files can be used. This change removes a specific preprocessing
step in the compilation, though the module only supports Fortran90
compilers. (Platform tested includes GNU, ALTIX, IBMAIX, and
IBMAIXMP.)
1.7.2 SQUANTM Parallel implementation
The module is parallelized, and a memory management is improved
when using QM/MM-Ewald summation method. The reciprocal K-space
terms require a huge memory allocation, and current modification
breaks down the memory allocation to each processor evenly but does
not require internode communication.
1.7.3 QM/MM Gaussian core-core interaction
To be consistent with other semiempirical QM/MM module in CHARMM,
Gaussian core-core interaction is introduced for AM1 and PM3 methods,
and a keyword is also introduced to turn off this interaction to be
compatible with older versions.
1.7.4 Bug fixes for MNDO97 and SQUANTM compilations.
Several bugs are fixed for a smooth compilation of MNDO97 and
SQUANTM module. In the MNDO97 module, the current fixes will allow
the use of GHO method as a default.
Files Added:
build/UNX/squantm.mk
source/squantm/sqnt_qm2_energy.src
source/squantm/sqnt_qm2_ewald.src
source/squantm/sqnt_qm2_gho.src
source/squantm/sqnt_qm2_grad.src
source/squantm/sqnt_qm2_mopac.src
source/squantm/sqnt_qmmm_int.src
source/squantm/sqnt_qmmm_module.src
source/squantm/sqnt_qmmm_util.src
source/fcm/qm2_array_locations.fcm
source/fcm/qm2_constants.fcm
source/fcm/qm2_conversions.fcm
source/fcm/qm2_double.fcm
source/fcm/qm2_elements.fcm
source/fcm/qm2_parameters.fcm
Files Removed:
build/UNX/squantm.mk_altix
build/UNX/squantm.mk_gnu
build/UNX/squantm.mk_ibmaix
build/UNX/squantm.mk_ibmaix64
source/squantm/sqnt_qm2_energy.src90
source/squantm/sqnt_qm2_ewald.src90
source/squantm/sqnt_qm2_gho.src90
source/squantm/sqnt_qm2_grad.src90
source/squantm/sqnt_qm2_mopac.src90
source/squantm/sqnt_qmmm_int.src90
source/squantm/sqnt_qmmm_module.src90
source/squantm/sqnt_qmmm_util.src90
source/squantm/assert.h
source/squantm/constants.h
source/squantm/conversions.h
source/squantm/dprec.h
source/squantm/qm2_array_locations.h
source/squantm/qm2_elements.h
source/squantm/qm2_parameters.h
Files Modified:
install.com
doc/qmmm.doc
doc/mndo97.doc
source/charmm/iniall.src
source/energy/energy.src
source/fcm/gamess.fcm
source/fcm/mndgho.fcm
source/mndint/mndo97q.tar.gz
source/squantm/sqnt_ene.src
source/squantm/sqnt_nbnd.src
source/squantm/sqnt_setup.src
------------------------------------------------------------------------------
1.8 Z-Module Updates and Various Enhancements
Developer : Robert J. Petrella
Email Address : petrella@fas.harvard.edu
Institution : Harvard University
Date : June 30, 2007
1.8.1 IRANdom random integer generator
IRANdom command--generates series of random numbers from uniform
distributions.
Files Modified:
doc/miscom.doc
source/charmm/miscom.src
source/util/random.src
..............................................................................
1.8.2 SECOnd comparison coordinate set
SECOnd comparison coordinate set for use in coordinate manipulation
commands
Prefx Keys: COMP2
Files Modified:
doc/corman.doc
source/fcm/coordc.fcm
source/gener/genpsf.src
source/io/mainio.src
source/manip/corman.src
..............................................................................
1.8.3 Z Module Update
ZMODule (structure prediction module) is refined and enhanced:
- compatibility with CMAP and Urey-Bradley terms added
- call to SUBROUTINE CODES implemented
- name changed from ZERO module
- other minor changes
Prefx Keys: ZEROM, ACTBOND
Files Modified:
doc/zerom.doc
source/charmm/charmm_main.src
source/fcm/actclus.fcm
source/zerom/zerom1.src
source/zerom/zerom2.src
..............................................................................
1.8.4 Bonded energy terms for active atoms
Facility for selecting atoms to be included in bonded energy
calculations.
Prefx Keys: ACTBOND
Files Modified:
source/charmm/iniall.src
source/energy/eintern.src
source/energy/enefscal.src
source/energy/energy.src
=========================================================================
1.9 Reading/Writing NAMD binary file in CHARMM
Developer : Andrei A. Golosov and Milan Hodoscek
Email Address : agolosov@tammy.harvard.edu
Institution : Harvard University
Date : June 30, 2007
CHARMM is now able to read and write NAMD (molecular dynamics program
from Klaus Schulten's lab) binary coordinates/velocities restart
files. To do this, READ_NAMD and WRITE_NAMD subroutines are called in
source/io/mainio.src These routines are written in C in
source/machdep/cstuff.c
Support files have been added for conversion of CHARMM psf file into
NAMD xplor psf file and creation of auxilarily pdb files for NAMD
necessary for simulations of solvated systems with fixed and
harmonically restrained solute.
Files Added:
support/namd/namdpsf.inp
support/namd/fix_solute.inp
support/namd/harm_rest_solute.inp
test/c35test/namdio.inp
Files Modified:
doc/io.doc
source/io/mainio.src
source/machdep/cstuff.c
------------------------------------------------------------------------------
1.10 Fast Analytical Continuum Treatment of Solvation
Developer : Urs Haberthuer and Amedeo Caflisch
Email Address : caflisch@bioc.uzh.ch
Institution : Department of Biochemistry, University of Zurich
Date : June 25, 2007
FACTS is an efficient method for calculating the free energy of
solvation of a (macro)molecule embedded in a continuum solvent
(Haberthuer and Caflisch, J. Comput. Chem. under review). It is based
on the fully analytical evaluation of the volume and spatial symmetry
of the solvent that is displaced from around a solute atom by its
neighboring atoms. The two measures of solvent displacement are
combined in empirical equations to approximate the atomic (or self)
electrostatic solvation energy and the solvent accessible surface area.
The former directly yields the effective Born radius, which is used in
the generalized Born formula to calculate the solvent-screened
electrostatic interaction energy. The solvent accessible surface area
is used to approximate the non-polar contribution to solvation. FACTS
is only four times slower than using the vacuum energy in molecular
dynamics simulations of peptides and proteins.
Files Added:
doc/facts.doc
source/fcm/facts.fcm
source/misc/fctall.src
test/c35test/facts.inp
Files Modified:
install.com
source/charmm/charmm_main.src
source/charmm/iniall.src
source/energy/energy.src
source/energy/eutil.src
source/energy/printe.src
source/fcm/energy.fcm
source/fcm/image.fcm
source/fcm/inbnd.fcm
source/nbonds/nbonds.src
source/nbonds/nbutil.src
------------------------------------------------------------------------------
1.11 RXMD: A Module to Switch PES during Dynamics Run
Developer : Markus Meuwly and Jonas Danielsson
Email Address :
Institution : University of Basel
Date : February 24, 2007
The Reactive Molecular Dynamics (RMD) method allows to do molecular
dynamics using two potential energy surfaces provided by the user, such that
the dynamics always takes place on the lowest surface. Crossings are detected
automatically and occur by a fast, smooth, switching centered in time at the
point where the crossing was detected. The implementation currently assumes
that the long-range interactions are handled with the SHIFT setting for
electrostatics and the SWITCH for the Lennard-Jones potential. To include
RMD in the compilation, the flag RMD should be included in pref.dat
Prefx Keys: RMD
Files Added:
doc/cross.doc
source/misc/cross.src
source/fcm/cross.fcm
source/fcm/saveforces.fcm
test/c35test/cross.inp
test/data/cross.par
test/data/hemeonly.psf
test/data/top_all22cross_prot.inp
test/data/par_all22cross_prot.inp
Files Modified:
source/charmm/iniall.src
source/charmm/miscom.src
source/dynamc/dcntrl.src
source/dynamc/dynamc.src
source/energy/energy.src
source/energy/eutil.src
source/fcm/energy.fcm
source/misc/genetic.src
source/pert/epert.src
source/pert/icpert.src
------------------------------------------------------------------------------
1.12 Karolinska Institute Fixes and Enhancements
Developer : Lennart Nilsson
Email Address : Lennart.Nilsson@biosci.ki.se
Institution : Karolinska Institute
Date : June 28, 2007
1.12.1 Fast nonbonded interactions using lookup tables
Fast (twofold speedup vs ENBAEXP) lookup tables implemented for
solvent-solvent, solvent-solute and solute-solute interactions. All nonbonded
schemes handled by ENBAEXP (including real space part of PME) are handled.
For now only three site water models can be used (TIP3P, SPC, SPC/E), but TIP4P
is being implemented. The code works in parallel and with peridic boundary
conditions. Non-bonded lists generated with BYGRoup or BYCB options work.
The code is protected by pref.dat keyword LOOKUP. Previous keywords (SLVSLV,
LNTAB1) are obsolete and have been removed in the supplied files.
[Note] There is a problem with the testcase when running in parallel on more
than 2 CPUs; part of the testcase uses a very small system and this causes
some list to be empty when the job is split over 4 CPUs. For real systems
there is no problem.
Files Added:
source/nbonds/elookup.src
source/nbonds/lookup_module.src
test/c35test/lookup.inp
Files Removed:
source/nbonds/ewwnb.src
test/c34test/nbsolv.inp
Files Modified:
doc/nbonds.doc
source/charmm/iniall.src
source/charmm/miscom.src
source/energy/energy.src
source/image/eimage.src
source/image/nbndgcm.src
source/image/upimag.src
source/nbonds/nbonds.src
source/nbonds/nbutil.src
source/nbonds/enbaexp.src
source/fcm/dimens.fcm
..............................................................................
1.12.2 Bugfixes
File modified
source/correl/solana.src Missing IREAL4 in call to FREHP
source/dynamc/dynio.src Constant in calls to extractint8, writint4,
source/util/datstr.src fillint4 replaced with variable
source/image/images.src Memory allocation for PBCUBES and IMCUBES fixed
------------------------------------------------------------------------------
1.13 Developments at the National Institute of Health
Developers : Jhih-Wei Chu, H. Lee Woodcock,
and Bernard R. Brooks
Email Address : jwchu@berkeley.edu, hlwood@nih.gov and brb@nih.gov
Institution : Laboratory of Computational Biology, NHLBI, NIH
Date : July 1, 2007
1.13.1 Holonomic Constraints for Order Parameters
Four order paramters have been implemendted.
1. The difference between two bond lengths
2. The distance between two atoms
3. Equal (rms)distances between two references
4. Equal (rms)distances between replicas of a reaction path.
Prefx Keys: RXNCONS
Files Added:
doc/rxncons.doc
source/fcm/rxncons.fcm
source/fcm/rxnconswt.fcm
source/fcm/rxncons3.fcm
source/fcm/rxncons4.fcm
source/manip/rxcons.src
source/manip/rxcons_1.src
source/manip/rxcons_2.src
source/manip/rxcons_3.src
source/manip/rxcons_4.src
test/c35test/rxncons.inp
test/c35test/rxncons_rpath.inp
Files Modified:
doc/commands.doc
source/charmm/charmm_main.src
source/charmm/iniall.src
source/dynamc/dcntrl.src
source/dynamc/dynamc.src
source/gener/update.src
source/manip/holonom.src
..............................................................................
1.13.2 Extensions for QM/MM and Reaction Path Methods
1. Full QM/MM Normal Mode Analysis via Q-Chem
2. Delocalized Gaussian Blurred MM charges support added via Q-Chem
3. Added Block support to SCC-DFTD replica path/NEB method
4. Clean up of SD minimizer (i.e. removal of unnecessary pathway method code)
5. Fixed NEB SD minimzer to decouple off-path and in-path forces during
minimizations
Prefx Keys: QCHEM, SCCDFTB, REPLICA, RPATH, GAMESS, GAMESS-UK
Files Added:
test/c35test/qm-mm_dgmm.inp
test/c35test/qm-mm_nmodes.inp
test/data/qchem_gpc.inp
test/data/qchem_pc.inp
test/data/qm-hess.inp
test/data/qmmm-hess.inp
Files Modified:
doc/gamess-uk.doc
doc/qchem.doc
doc/replica.doc
doc/sccdftb.doc
source/charmm/iniall.src
source/energy/energy.src
source/fcm/gamess.fcm
source/gukint/gukini.src
source/minmiz/steepd.src
source/sccdftbint/sccdftbini.src
------------------------------------------------------------------------------
1.14 Developments at NIC-Slovenia
Developer : Milan Hodoscek
Email Address : milan@cmm.ki.si
Institutions : National Institutes of Chemistry, Ljubljana, Slovenia
and National Institute of Health, Bethesda, MD, USA
Date : June 30, 2007
1.14.1 GAMESS/CHARMM QM/MM interface: support for the latest GAMESS
Support for the latest version of GAMESS (24 MAR 2007 R1) was
added to CHARMM. New development in this version includes: TD-DFT
method, many new functionals were added totaling to almost 50, another
MP2 parallelization, etc. Also the interface code uses XXLARGE charmm
as default.
Files Modified:
build/UNX/gamess.mk
build/UNX/gmscomp
doc/gamess.doc
doc/diesel.doc
source/fcm/gamess.fcm
source/gamint/ddi.src
source/gukint/gukini.src
.............................................................................
1.14.2 REPDSTR
The REPlica DiSTRibuted functionality was further extended:
parallel IF, STREam, OUTU is also working now.
Files Added:
test/data/ala2.crd_0
test/data/ala2.crd_1
test/data/ala2.crd_2
test/data/ala2.crd_3
Files Modified:
doc/repdstr.doc
doc/replica.doc
source/charmm/charmm_main.src
source/charmm/iniall.src
source/charmm/miscom.src
source/energy/energy.src
source/energy/epath.src
source/fcm/repdstr.fcm
source/fcm/stream.fcm
source/gener/repdstr.src
source/io/mainio.src
source/machdep/paral1.src
source/util/parse.src
test/c34test/rexc.inp
.............................................................................
1.14.3 Multi Scale Command: MSCALE
The multiscale command causes CHARMM to run several
independent but connected calculations using subsystems. The
calculations can use either CHARMM or other programs with a consistent
interface. The suppoted interface programs are implemented for the
following ab initio packages: NWCHEM, GAUSSIAN, MOLPRO.
Files Added:
doc/mscale.doc
build/UNX/mscale.mk
source/mscale/mscale.src
source/mscale/nwchem.f03
source/mscale/Makefile
test/c35test/mscale.inp
test/c35test/sys1.inp
test/c35test/sys2.inp
test/c35test/sys3.inp
Files modified:
install.com
build/UNX/Makefile_gnu
build/UNX/energy.mk
source/charmm/charmm_main.src
source/charmm/iniall.src
source/energy/energy.src
source/machdep/machio.src
source/machdep/startup.src
source/pert/epert.src
.............................................................................
1.14.4 Torsional Cross Term: CMAP
The CMAP code was extended to support REPLICA and BLOCK
Files modified:
source/energy/ecmap.src
source/fcm/block.fcm
source/gener/replica.src
source/gener/update.src
source/pert/block.src
.............................................................................
1.14.5 Small fixes:
Files modified:
source/cff/enbfast_cff.src (##IMCUBES vs. ##PBCUBES)
source/cff/ewald_cff.src (##IMCUBES vs. ##PBCUBES)
source/correl/rdfsol.src (SOLPRT() -> WSOLPRT() conflict with GAMESS)
source/dynamc/dcntrl.src (##LDM added to compile ##PERT without it)
source/energy/energy.src (##BANBA added to ##IF)
source/machdep/cstuf.c (OpenMPI fix)
source/machdep/startup.src (##PARCMD fix)
source/manip/corman.src (write(*,*) fix)
source/misc/mmfp.src (INTEGER(4) vs INTEGER(8) problem)
source/misc/scpism.src (DERVIS() -> SCDERIVS() conflict with GAMESS)
source/misc/pbeq2.src (ADDINT() -> PBADDINT() conflict with GAMESS)
source/nbonds/pblist.src (##FORTRAN90 - deprecated)
test/c33test/hqbm_single.inp (old integrator not parallel!)
test/c34test/vmodwat.inp (vibran and quasi not parallel!)
test/c34test/testzero.inp (if ?zerom was missing)
------------------------------------------------------------------------------
1.15 Block and QM/MM Normal Mode Updates
Developer : Qiang Cui
Email Address : cui@chem.wisc.edu
Institution : University of Wisconsin - Madison
Date : July 2, 2007
1.15.1 Block normal mode, QM/MM normal mode
Compilation problems with the VIBBLOCK flag are fixed. A problem that
QM/MM NMA with REDU FIX FINITE doesn't work for large systems is fixed.
Files Modified:
source/misc/hbuild.src HSBLC to 0 fix for !##VIBBLOCK
source/misc/fitchrg.src Add protection (##IFN VIBBLOCK) for PYTHAG
source/energy/hqbm.src Misspelling VIBBLOCK as VIBBLOVK (2 times)
source/vibran/vibblock.src Remove the definition/references of ATOMON
source/misc/eef1.src
..............................................................................
1.15.2 For QM/MM NMA
In the section that processes 'REDU',somehow some key changes
(the avoid of allocating the hessian for the entire system even when many
atoms are fixed) have been taken out. QC put back in.
Files Modified:
source/vibran/vibran.src
------------------------------------------------------------------------------
1.16 Developments at University of Vienna
Developers : Stefan Boresch
Email Address : stefan@mdy.univie.ac.at
Institution : University of Vienna
Date : May 7, 2007
This fix addresses two distinct problems:
I. Having a lot of nonbonded exclusions in a RESI entry in the RTF (e.g.
useful to simulate dual topology with PERT) can result in a buffer
overflow in SUBROUTINE MAKINB (nbexcl.src). NOTE: The bug was discovered
in connection with PERT, but is completely independent of any PERT use!
II. Switching extended electrostatics on and off while PERT is
activated causes crashes.
Files involved:
source/pert/pert.src
source/nbonds/nbexcl.src
source/nbonds/nbonds.src
source/nbonds/nbutil.src
------------------------------------------------------------------------------
1.17 Florida State University Developments
Developers : Wei Yang
Email Address : yang@sb.fsu.edu
Institution : Florida State University, School of Computational Science
Date : July 3, 2007
1.17.1 Adaptive integration (ADIN) method for hybrid MD/MC simulation
Adaptive integration method is implemented based on in the original
chemical space hybrid MD/MC simulation, which has the acceptance probability:
P_accpt[lambda0->lambda1] = min{1,exp[-beta*(lambda1-lambda0)*dUdl(lambda0)]}.
In the new method, an adaptive reweighting scheme was realized based on the
following equation:
min{1,exp[-beta*(lambda1-lambda0)*dUdl(lambda0)]
P_accpt[lambda0->lambda1] = --------------------------------------------------
exp[-beta*(+)/2]} ,
where dUdl denotes energy derivative with respect to each visiting lambda
and <> represents ensemble average.
Files Added:
source/fcm/blockappend.fcm
test/c35test/test/adin.inp
Files Modified:
doc/block.doc
source/pert/block.src
source/dynamc/dynamc.src
..............................................................................
1.17.2 Theta-dynamics
This is an alternative method for the original lambda-dynamics. Lambda**2
is replaced by sin(theta)**2 and (1-lambda**2) by cos(theta)**2. Theta, instead
of lambda, now is the variable for propagation. This implementation can avoid
the artifacts brought in by the constant external works in the Lagarangian
Multiplier boundary treatment.
Files Added:
test/c35test/test/thetadm.inp
Files Modified:
doc/block.doc
source/energy/energy.src
source/dynamc/dcntrl.src
source/dynamc/dynamc.src
source/dynamc/dynio.src
source/dynamc/dynlng.src
source/pert/block.src
------------------------------------------------------------------------------
1.18 Bugfixes and Enhancements
1.18.1 ifort v8.0 compilation
Developer : Robert Best
Email Address : r.b.best.00@cantab.net
Institution : NIH
Date : March 9, 2007
CHARMM will not compile with ifort v8.0 due to the variable N used prior
to declaration in pme.src, subroutine CFFTF. Declaration of n is switched to
before declaration of wsave(4*n).
Files involved:
source/nbonds/pme.src
..............................................................................
1.18.2 Parallelization of the SCP-ISM
Developer : Sergio A. Hassan
Email Address : mago@helix.nih.gov
Institution : CMM/DCB National Institutes of Health
Date : June 20, 2007
The SCP-ISM (Screened Coulomb Potentials-Implicit Solvent Model) was
parallelized using standard MPI routines coded into the scpism.src.
Files Modified:
doc/scpism.doc
source/fcm/scpism.fcm
source/misc/scpism.src
..............................................................................
1.18.3 X11 color map
Developer : Rick Venable
Email Address : venabler@nhlbi.nih.gov
Institution : NIH/NHLBI
Date : July 6, 2007
Colors were not displayed properly for the X11 screen graphics.
Declare INTEGER*4 COLOR_MAP instead of INTEGER.
[Note] Promotion of all INTEGER variables to INTEGER*8 led to a word length
mismatch for the COLOR_MAP array, which is passed to a C routine.
Files involved:
source/fcm/graph.fcm
The following are introduced and modified during the development period
of August 15, 2007 through February 15, 2008. The developments are based on
CHARMM version c35a1. Some enhancements and fixes are also incorporated into
c34b2 as described in the c34b1 release note (ChangeLog.c34).
2.1 Developments at the National Institute of Health
Developer : H. Lee Woodcock, Xiongwu Wu, Richard W. Pastor and
Bernard R. Brooks
Email Address : wuxw@nhlbi.nih.gov, hlwood@nih.gov and brb@nih.gov
Institution : Laboratory of Computational Biology, NHLBI, NIH
Date : December 28, 2007
2.1.1 Universal IPS method for anisotropic systems (XW)
The IPS method is extended to anisotropic simualtion systems. The extended
IPS method calculates long-range interactions in two steps. The first step
calculates the interaction within the local region defined by CTOFNB or RIPS.
The second step calculates the difference of an anisotropic system (defined by
radius RAIPS, which is set by default the diagonal distance of the PBC box).
The first step is done the same way as the cutoff methods by summing over local
atom pairs. The second step is done through the convolution thereom which can
be efficiently calculated using the FFT technique. This method can be used for
both homogenouse and hetergenous systems, as well as non-periodic systems like
in vacuume. For interfacial systems, 1+2D IPS can be used for the second step
calculation. Also, the IPS method can be applied for L-J interaction alone to
be combined with Ewald method for charge interaction calculation.
This extension make the IPS method a convenient way for any type of simulation
study.
Prefx Keys: NBIPS
Files Modified:
install.com
build/UNX/energy.mk
build/UNX/Makefile_sgi
build/UNX/Makefile_sgi64
doc/nbonds.doc
source/energy/energy.src
source/fcm/nbips.fcm
source/image/nbondm.src
source/machdep/startup.src
source/misc/fctall.src
source/nbonds/enbaexp.src
source/nbonds/enbfast.src
source/nbonds/enbips.src
source/nbonds/enbond.src
source/nbonds/enbonda.src
source/nbonds/enbondg.src
source/nbonds/nbonda.src
source/nbonds/nbondg.src
source/nbonds/nbonds.src
source/nbonds/nbutil.src
source/util/new_timer.src
source/vibran/vibran.src
test/c32test/ips_tip216.inp
..............................................................................
2.1.2 Extensions for QM/MM and Reaction Path Methods (HLW)
1. Full QM/MM Normal Mode Analysis via Q-Chem (can also save and read
previously computed Hessians)
2. Delocalized Gaussian Blurred MM charges support added via Q-Chem (added
testcase)
3. Micro-iteration optimization procedure for QM/MM and QM/MM Replica Path
(not fully tested)
4. Can use Q-Chem's internal coordinate optimization with QM/MM interface
5. Can place Lennard-Jones potentials on MM atoms during QM/MM optimizations
using Q-Chem's optimizer (useful for preventing over-polarization)
6. Added ability to reuse saved orbitals between energy calls and/or
optimizations
7. Added ability to read in and set charges on the QM region of a QM/MM
calculation (useful for speeding up QM/MM optimizations)
8. Improved lone pair handling in FINDEL
Prefx Keys: QCHEM, REPLICA, RPATH, VIBRAN
Files Added:
test/c35test/qmmm_dgmm.inp
test/c35test/qmmm_normal_modes.inp
test/data/qchem_gpc.inp
test/data/qchem_pc.inp
test/data/qm-hess.inp
test/data/qmmm-hess.inp
Files Modified:
doc/qchem.doc
source/charmm/iniall.src
source/energy/energy.src
source/fcm/gamess.fcm
source/gukint/gukini.src
source/quantum/addlnat.src
test/c33test/ts_opt_cyclohexane.inp
..............................................................................
2.1.3 Vibrational Subsystem Analysis
Added the VSA method. The VSA method will be useful in a variety of situation;
1. examination of local-global motion
2. performing accurate NMA while eliminating unwanted degrees of freedom
3. eliminating excess noise from large NMA (i.e. QM/MM)
4. performing NMA while not at a stationary point with respect to all degrees
of freedom
5. integration of light particle during NMA (i.e. application to polarizable
models)
Prefx Keys: QCHEM, VIBRAN
Files Added:
test/c35test/vsa_butane.inp
Files Modified:
source/vibran/vibran.src
source/vibran/vibsub.src
source/manip/scalar.src
------------------------------------------------------------------------------
2.2 Developments at NIC-Slovenia
Developer : Milan Hodoscek
Email Address : milan@cmm.ki.si
Institutions : National Institutes of Chemistry, Ljubljana, Slovenia
and National Institute of Health, Bethesda, MD, USA
Date : January 2, 2008
2.2.1 Add new parameter to the PRINTE subroutine
PRINTE calculates the GRMS value internally. However in some
cases this is not correct since the printing process gets the value
from the other processes. See for example source/csa/csacomm.src. A
new flag QGRMS was introduced into parameter list to skip the GRMS
calculation.
Files modified:
source/energy/printe.src
source/dynamc/dynamcv.src
source/dynamc/dynamc4.src
source/dynamc/dynamc.src
source/dimb/nmdimb.src
source/dynamc/dynamln.src
source/dynamc/dynamvv.src
source/dynamc/dynamvv2.src
source/dynamc/tamd.src
source/dynamc/tmd.src
source/energy/ediff.src
source/energy/eutil.src
source/energy/intere.src
source/image/xtlfrq.src
source/io/mainio.src
source/mbond/mbback.src
source/mc/mcener.src
source/minmiz/abner.src
source/minmiz/conjug.src
source/minmiz/egrad1.src
source/minmiz/nraph.src
source/minmiz/powell.src
source/minmiz/steepd.src
source/minmiz/tndriv.src
source/misc/genetic.src
source/misc/genetic2.src
source/pert/block.src
source/pert/epert.src
source/pert/pert.src
source/rxncor/path.src
source/vibran/redbas.src
source/vibran/vibcom.src
source/vibran/vibio.src
source/vibran/vibsub.src
source/zerom/zerom2.src
..............................................................................
2.2.2 Conversion of epath.src into the new fortran standard
RPATH module is converted to the new fortran standard.
No stack or heap usage in the rpath.src file.
Files Modified:
source/charmm/charmm_main.src
source/charmm/iniall.src
source/charmm/miscom.src
source/dynamc/dcntrl.src
source/energy/energy.src
source/energy/epath.src
source/fcm/path.fcm
source/gener/repdstr.src
source/manip/rxcons_4.src
source/minmiz/abner.src
source/util/parse.src
..............................................................................
2.2.3 FINDEL routine not protected by QM/MM flags
FINDEL routine is needed in other than QM codes
file modified:
source/quantum/addlnat.src
..............................................................................
2.2.4 Increased the size of MAXPAD, MAXSHK in dimens.fcm
Increased dimensions so that water system with near 25000 atoms can be run
in medium sized charmm.
Modified file:
source/fcm/dimens.fcm
.......................................................
2.2.5 Unification of libraries in Makefile_ files
Fixed the inconsistencies in the build/UNX/Makefile_ files. In
order to avoid editing them separately the objlibs.mk include file is
created. install.csh is also modified accordingly.
Files Added:
build/UNX/objlibs.mk
Files Modified:
build/UNX/Makefile_alpha
build/UNX/Makefile_alphamp
build/UNX/Makefile_altix
build/UNX/Makefile_em64t
build/UNX/Makefile_g95
build/UNX/Makefile_gnu
build/UNX/Makefile_hpitanium
build/UNX/Makefile_hpux
build/UNX/Makefile_ibmaix
build/UNX/Makefile_ibmaix64
build/UNX/Makefile_ibmaixmp
build/UNX/Makefile_ibmaixmp64
build/UNX/Makefile_ibmlnxmp
build/UNX/Makefile_ibmlnxmp_so
build/UNX/Makefile_itanium
build/UNX/Makefile_osx
build/UNX/Makefile_sgi
build/UNX/Makefile_sgi64
build/UNX/Makefile_sun
build/UNX/Makefile_sun64
build/UNX/Makefile_sunmpi
build/UNX/Makefile_t3e
------------------------------------------------------------------------------
2.3 Dynamic Importance Sampling (DIMS)
Developers : Juan R. Perilla, Oliver Beckstein, Anu Nagarajan,
Jennifer Johnston, Elizabeth Denning, Tom Woolf
Email Address : jrperillaj@jhu.edu, orbeckst@jhmi.edu, anu@jhu.edu,
jen.johnston@jhmi.edu, edennin3@jhu.edu, twoolf@jhmi.edu
Institution : Johns Hopkins University, School of Medicine
Date : Dec 19, 2007
Dynamic Importance Sampling (DIMS) is a method that generates ensembles of
transitions between a given initial and final state. Typically, those states
are experimental structures in two different functional states. What sets DIMS
apart from other methods is that no reaction coordinate needs to be defined in
advance and that the quality of a transition can be assessed with a score
during the simulation.
DIMS has a rather large number of options implemented and it can access
several other parameters from other functions that it uses. DIMS can use
different 'flavors' to bias the transition. (We use 'flavors' in favor of
'modes' in order to avoid confusion with normal modes.). DIMS-Block Normal
Modes goes from the origin structure towards the target structure by
displacing atoms on the conformational space, using collective motion
information---the normal modes---as bias. Two additional features increase
the quality and diversity of trajectories: linear combinations of normal
modes and mode self-avoidance. Another flavor, DIMS-Cartesian, generates
targeted stochastic trajectories by using a soft-ratcheting algorithm.
Prefx Keys: DIMS
VIBBLOCK (required)
Files Added:
doc/dims.doc
source/dynamc/dims.src
source/fcm/dims.fcm
Files Modified:
build/UNX/dynamc.mk
doc/corman.doc
doc/dynamc.doc
doc/vibran.doc
doc/commands.doc
doc/correl.doc
doc/io.doc
source/charmm/charmm_main.src
source/charmm/iniall.src
source/correl/anacor.src
source/dynamc/dcntrl.src
source/dynamc/dynamc.src
source/fcm/new_time.fcm
source/io/mainio.src
source/manip/corman.src
source/util/new_timer.src
test/c35test/dims.inp
test/c35test/dims_nm.inp
test/data/1bdc_dims_start.pdb
test/data/1bdc_dims_targ.pdb
------------------------------------------------------------------------------
2.4 FACTS Improvements
Developer : Urs Haberthuer, Francois Marchand and Amedeo Caflisch
Email Address : caflisch@bioc.uzh.ch
Institution : Department of Biochemistry, University of Zurich
Date : December 20, 2007
1. The FACTS (Fast Analytical Continuum Treatment of Solvation) module
was parallelized using the standard CHARMM communication routines
coded in the paral1.src.
2. Van der Waals dispersion term was added to FACTS according to
E. Gallicchio and RM Levy, J Comput Chem 25: 479--499, 2004.
3. Minor improvements: timer added, better memory management, code layout.
Prefx Keys: FACTS
Files Added:
test/c35test/facts_p19.inp
test/c35test/facts_p22.inp
Files Removed:
test/c35test/facts.inp
Files Modified:
build/UNX/misc.mk
doc/facts.doc
doc/parallel.doc
doc/preflx_list.doc
source/fcm/facts.fcm
source/energy/energy.src
source/image/images.src
source/misc/fctall.src
source/nbonds/nbonds.src
source/nbonds/nbutil.src
source/util/new_timer.src
------------------------------------------------------------------------------
2.5 Crystal Simulation Enhancements
Developer : Guillaume Lamoureux, Pedro E. M. Lopes and
Alexander D. MacKerell, Jr.
Email Address : glamoure@alcor.concordia.ca, lopes@outerbanks.umaryland.edu
alex@outerbanks.umaryland.edu
Institution : Concordia University and University of Maryland, Baltimore
Date : December 31, 2007
Allow crystal simulations of different symmetries using the VV2
integrator with the polarizable Drude model. The implemented changes have the
purpose of extending the constant-pressure code for the VV2 integrator.
Extensions are added to perform crystal simulations in different symmetries.
Additions have been made in the TPCONTROL command as shown below.
Two new flags in the BARO line of TPCONTROL:
ZONLY: To allow scaling along the Z-direction only.
FULL: To allow scaling of all degrees of freedom of the crystal type,
without breaking its symmetry.
Note that the code does not work with the MONOCLINIC lattice type.
Files Modified:
doc/tpcntrl.doc
source/fcm/nose.fcm
source/dynamc/dcntrl.src
source/dynamc/tpcontrol.src
source/dynamc/dynamvv2.src
source/dynamc/dynio.src
------------------------------------------------------------------------------
2.6 FITPARAM: Generic Parameter Optimization Module
Developer : Victor Anisimov, Alexander MacKerell
Email Address : victor@outerbanks.umaryland.edu
Institution : University of Maryland at Baltimore
Date : December 31, 2007
New fitting procedure (FITPARAM) is implemented as a part of the
FITCHARGE module. FITPARAM is a parameter fitting procedure that is primarily
designed to fit partial atomic charges and atomic polarziabilities based on
the Drude oscillator model to interaction energy data and dipole moments,
though it may be applied to fitting of other parameters. It supports
optimization of multiple parameters for a series of model compounds sharing
common parameter sets. Different weights can be assigned to different target
data. Optimized parameters can be restrained to their corresponding initial
values by using a parabolic penalty function. FITPARAM performs non-linear
least square fitting using the Levenberg-Marquardt algorithm.
A revision of FITCHARGE has also been made to reduce its memory requirement.
Prefx Keys: FITCHG
Files Added:
doc/fitparam.doc
test/c35test/antoine.inp
test/data/antoine.ini
test/data/antoine.exp
Files Modified:
source/misc/fitcharge.src
source/charmm/charmm_main.src
------------------------------------------------------------------------------
2.7 Dyanmic Memory Usage for RMD
Developer : Stephan Lutz, Markus Meuwly
Email Address : stephan.lutz@unibas.ch, m.meuwly@unibas.ch
Institution : Department of Physical Chemistry, University of Basel
Date : December 18, 2007
Solved the out-of-memory compilation error when compiling with larger
charmm-sizes and prefix RMD. Charmm-size dependent common block arrays in
cross.fcm and saveforces.fcm were replaced by dynamical allocatable Fortran90
arrays. Memory for these arrays only gets allocated by giving the RXMD command
and for some of these arrays additionally the DYNAmics command.
Files removed:
source/fcm/saveforces.fcm
Files modified:
build/UNX/charmm.mk
build/UNX/dynamc.mk
build/UNX/energy.mk
build/UNX/misc.mk
build/UNX/pert.mk
doc/cross.doc
source/charmm/miscom.src
source/fcm/cross.fcm
source/dynamc/dnctrl.src
source/dynamc/dynamc.src
source/energy/energy.src
source/misc/cross.src
source/pert/epert.src
source/pert/icpert.src
------------------------------------------------------------------------------
2.8 Simulated Scaling
Developer : Wei Yang
Email Address : yang@mailer.sb.fsu.edu
Institution : Florida State University
Date : January 31, 2008
This is the method for the localized enhanced sampling and simultaneous
alchemical free energy simulations. The recursion in Wang-Landau algorithm
is adopted in this method. For more details, please check Li, H., Fajer, M.,
and Yang, W. 2007. Simulated scaling method for efficient localized
conformational sampling and simultaneous alchemical free energy simulation:
A general method for MM, QM, and QM/MM simulations. J. Chem. Phys. 126:024106.
Files Added:
test/c35test/sstest.inp
Files Modified:
doc/block.doc
source/charmm/iniall.src
source/dynamc/dcntrl.src
source/dynamc/dynamc.src
source/fcm/blockappend.fcm
source/fcm/lambda.fcm
source/mc/mvhmc.src
source/pert/block.src
------------------------------------------------------------------------------
2.9 Karolinska Institute Fixes and Enhancements
Name : Lennart Nilsson
Email Address : Lennart.Nilsson@biosci.ki.se
Institution : Karolinska institutet
Date : December 28, 2007
(1) FORMAT correction and misunderstanding of insertion codes when reading
sequence from PDB file
Files involved:
source/io/coorio.src
source/io/psfres.src
(2) Changed internal variables to REAL*8 to allow large range.
Files involved:
source/util/calc.src
(3) Added option to supply radii for excluded volume correction in WMAIN.
Also added the excluded volume correction and the uncorrected g(r) to the
output. Changed an EQUIVALENCE statement that caused F90 compilation problems
Files involved:
doc/corman.doc
source/correl/solana.src
(4) Documented the NAME keyword for READ/WRITE commands
Files involved:
doc/io.doc
(5) The lookup command is available in c34b1.
Files involved:
test/c34test/lookup.inp NEW
test/c35test/lookup.inp DELETED
(6) Improved energy reporting when using LOOKUP NOENERGY
With the NOENERGY option the LOOKUP code only calculates the energies when
the non-bond list is updated. For image energies intermittently stored energy
was not properly reset, and hence reported energies could show jumps (no effect
on the dynamics though since forces are not affected).
Files involved:
source/image/eimage.src
------------------------------------------------------------------------------
2.10 Genetic Neural Network for QSPR
Developer : Jie Hu and Aaron Dinner
Email Address : dinner@uchicago.edu
Institution : The University of Chicago
Date : January 15, 2008
Artificial neural networks are used to fit data non-linearly (with possible
cross-validation) and either one of two genetic algorithms or exhaustive
enumeration can be used to select between the combinations of descriptors
that yield the best predictions for the target data. Input format and options
are described in gnn.doc.
Prefx Keys: GNN
Files Added:
doc/gnn.doc
source/fcm/gnn.fcm
source/misc/gnn.src
test/c35test/gnn.inp
test/data/gnn.dat
Files Modified:
build/UNX/misc.mk
source/charmm/charmm_main.src
------------------------------------------------------------------------------
2.11 Bugfixes
2.11.1 Bug fixed for trace-statistics collection routine in umbrella sampling
Name : Yousung Jung
Email Address : yousung@caltech.edu
Institution : California Institute of Technology
Date : September 11, 2007
CHARMM version: c33b2
In the present code (rxnene.src) of umbrella sampling, one can have more
'hits' in the .stt file than the number of MD steps actually taken, because
of the way the statistics is collected currently using the variable 'rxncnt'.
This would then yiled the free energy that is incorrect. This problem is
described also in rxnene.src.
In the fixed code, I use the actual MD step (MDSTEP2 = ISTEP) as a
control variable for updating the trace-statistics.
Files involved:
source/charmm/iniall.src
source/dynamc/dynamc.src
source/fcm/rxncom.fcm
source/rxncor/rxnene.src
source/rxncor/rxndef.src
..............................................................................
2.11.2 CLK_TCK Definition fixed
Developer : Mikhail Wolfson
Email Address : wolfsonm@mit.edu
Institution : MIT
Date : January 29, 2008
In the file source/machdep/cstuff.c, lines 391 - 393 read:
#if defined(CLOCKS_PER_SEC) /* fix for glibc-2.4 */
#define CLK_TCK CLOCKS_PER_SEC
#endif
This is incorrect. CLK_TCK and CLOCKS_PER_SEC do not have the same value on
many systems. It is fixed to
#if !defined(CLK_TCK) /* fix for glibc-2.4 */
extern long int __sysconf (int);
# define CLK_TCK ((__clock_t) __sysconf (2)) /* 2 is _SC_CLK_TCK */
#endif
Files involved:
source/machdep/cstuf.c
..............................................................................
2.11.3 Calling argument list fix (charmm.org forum)
tmdrhof removed from the calling argument list for shiftz
Files involved:
source/dynamc/dynamc.src
..............................................................................
2.11.4 Preprocessor directive fixes
Misplaced and missing preprocessor directives fixed (charmm.org forum)
Files involved:
source/energy/energy.src
source/energy/gbmvmodule.src
source/energy/genborn.src
source/misc/scpism.src
..............................................................................
2.11.5 NDEGF fix for VV2 Integrator
Developer : Ed Harder
Email Address : eharder@uchicago.edu
Institution : University of Chicago
Date : Feb 11, 2008
In the VV2 integrators the evaluation of the number of degrees of freedom
for the system is fixed. 6 degrees of freedom were erroneously being
subtracted for the VV2 integrator in source/dynamc/dcntrl.src. The result is
an incorrect assignment of kinetic energy to the system when using the Nose
thermostats with the VV2 integrator. The problem is only significant when 6 is
on the order of the total number of degrees of freedom in the system such as
a single molecule in vacuum. This is corrected and 3 degrees of freedom are
necessarily subtracted for the translational degrees of freedom lost when
the center of mass kinetic energy of the system is dissipated.
Files involved:
source/dynamc/dcntrl.src
..............................................................................
2.11.6 SCCDFTB bugs fixed
Developers : Qiang Cui
Email Address : cui@chem.wisc.edu
Institution : University of Wisconsin, Madson
Date : October 27, 2007
Free HEAP properly and initialize NSCCRP when SCCDFTB in use.
source/misc/gsbp.src
source/misc/pbeq.src
..............................................................................
2.11.7 MOLVIB document updated
Developer : Kenno Vanommeslaeghe
Email Address : kenno@outerbanks.umaryland.edu
Institution : University of Maryland, Baltimore
Date : January 8, 2008
Ducument the definition of linear bend internal coordinates in MOLVIB.
Files involved:
doc/molvib.doc
------------------------------------------------------------------------------
2.12 Force Field Parameter Updates
Developers : Alex D. MacKerell
Email Address : alex@outerbanks.umaryland.edu
Institution : University of Maryland, School of Pharmacy
Date : December 31, 2007
1) New carbohydrate force field parameters added. This includes an
extensive set of monosaccharides, acyclic analogs and glycosyl
linkages. The parameters are designed to be compatible with the
CHARMM22 and 27 additive force fields.
2) Lipids: Minor corrections to several residues used for generation
of lipids. Myristic acid added to allow for creation of DMPC. New
model compounds added to toppar_all27_lipid_model.str.
3) Proteins: Neutral N- and C-terminal patches added: NNEU and CNEU.
4) Proteins: Topology and parameter files WITHOUT CMAP no longer the
default and not included in the toppar directory. Note that the
original CHARMM22 model may be reproduced by including the SKIPE CMAP
command in the charmm input script.
5) Redundant toppar_history files deleted.
Files Added:
toppar/top_all35_sugar.rtf
toppar/par_all35_sugar.prm
Files Removed:
toppar/top_all22_prot_cmap.inp
toppar/par_all22_prot_cmap.inp
toppar/toppar_history/c32b2/toppar_history directory
toppar/toppar_history/c33b2/toppar_history directory
Files Modified:
doc/parmfile.doc
doc/rtop.doc
toppar/par_all22_prot.inp
toppar/par_all27_lipid.prm
toppar/par_all27_prot_lipid.prm
toppar/par_all27_prot_na.prm
toppar/stream/toppar_all22_prot_fluoro_alkanes.str
toppar/stream/toppar_all22_prot_model.str
toppar/stream/toppar_all27_lipid_cholesterol.str
toppar/stream/toppar_all27_lipid_model.str
toppar/stream/toppar_all27_na_model.str
toppar/top_all22_prot.inp
toppar/top_all27_lipid.rtf
toppar/top_all27_na_lipid.rtf
toppar/top_all27_prot_lipid.rtf
toppar/top_all27_prot_na.rtf
toppar/toppar_all.history
