parameterize
is a molecular force field parameterization tool.
Commonly used AMBER and CHARMM force fields contain parameters for biomolecules (proteins, nucleotides, saccharides, lipids, etc.), but lack parameters for other biologically relevant molecules (co-factors, drugs, etc.).
GAFF and CGenFF address this by using empirical rules and pre-computed data sets to derive parameters for arbitrary organic molecules. However, these parameters are not guaranteed to be transferable to all possible chemical environments.
parameterize
improves the quality of the parameters by using QM data, i.e. refitting ESP charges and rotatable
dihedral angle parameters. It fundamentally solves the problem of transferability.
Contents:
Capabilities:
AMBER
CHARMM
GAFF and GAFF2 (for AMBER)
CGenFF (for CHARMM)
Psi4
TeraChem (experimental)
Fitted to reproduce the electrostatic potential (ESP) of QM
Selected charges can be fixed during the fitting
Automatic rotatable dihedral detection
Automatic identical dihedral detection
Dihedral scanning with and without QM structure optimization
Parameter fitting to reproduce QM energies
Local machine
LSF/Slurm queuing systems
parameterize
takes a MOL2 file, parameterizes it, and outputs force field files in CHARMM and
AMBER formats. There are many options that are shown in detail below. The structure inside the output directory
(controlled by the --output
flag) is the following:
.
├── dihedral-opt
├── esp
├── minimize
└── parameters
The output inside dihedral-opt/
, esp/
, and minimize/
is related with the QM calculations performed during
the parameterization of the molecule and work as check-points for the different steps of the parameterization process.
The parameters
is where the relevant outputs are written with following format:
parameters/<force-field>/<theory>-<basis-set>-<vacuum/water>/
. Inside this directory there are the force-field
parameter files, as well as a plots/
directory, where one can check the optimization carried on by the tool.
parameterize
can be installed with Conda package manager:
Install Conda, if not yet available. We recommend to install the latest Miniconda following these instructions.
Install parameterize
into an active (default) Conda environment:
$ conda install -c acellera -c psi4 -c conda-forge parameterize
Alternative one, can install
parameterize
into a separe Conda environment:$ conda create -n param_env -c acellera -c psi4 -c conda-forge parameterize $ conda activate param_env
Acellera small molecule parameterization tool
usage: parameterize [-h] [-c CHARGE] [-l] [--rtf-prm <filename> <filename>]
[-ff {CGenFF_2b6,GAFF,GAFF2}]
[--fix-charge <atom name> [<atom name> ...]]
[-d A1-A2-A3-A4 [A1-A2-A3-A4 ...]]
[--code {Psi4,Gaussian}] [--theory {HF,B3LYP,wB97X-D}]
[--basis {6-31G*,6-31+G*,6-311G**,6-311++G**,cc-pVDZ,aug-cc-pVDZ}]
[--environment {vacuum,PCM}] [--min-type {None,qm,mm}]
[--charge-type {None,Gasteiger,AM1-BCC,ESP}] [--no-dihed]
[--scan-type {None,qm,mm}]
[--dihed-num-iterations DIHED_NUM_ITERATIONS]
[--dihed-fit-type {iterative,NRS}] [-q {local,Slurm,LSF}]
[-n NCPUS] [-m MEMORY] [-o OUTDIR] [--seed SEED]
[--version]
filename
Molecule file in MOL2 format
Total charge of the molecule (default: sum of partial charges)
List parameterizable dihedral angles
CHARMM RTF and PRM files
Possible choices: CGenFF_2b6, GAFF, GAFF2
Initial atom type and parameter assignment (default: “GAFF2”)
Fix atomic charge during charge fitting (default: none)
Select dihedral angle to parameterize (default: all parameterizable dihedral angles)
Possible choices: Psi4, Gaussian
QM code (default: “Psi4”)
Possible choices: HF, B3LYP, wB97X-D
QM level of theory (default: “wB97X-D”)
Possible choices: 6-31G*, 6-31+G*, 6-311G**, 6-311++G**, cc-pVDZ, aug-cc-pVDZ
QM basis set (default: “6-311++G**”)
Possible choices: vacuum, PCM
QM environment (default: “vacuum”)
Possible choices: None, qm, mm
Type of initial structure optimization (default: “mm”)
Possible choices: None, Gasteiger, AM1-BCC, ESP
Partial atomic charge type (default: “AM1-BCC”)
Do not perform QM scanning of dihedral angles
Possible choices: None, qm, mm
Type of structure optimization when scanning dihedral angles (default: “mm”)
Number of iterations during the dihedral parameter fitting
Possible choices: iterative, NRS
Dihedral fitting method. Can be either iterative or naive random search (NRS).
Possible choices: local, Slurm, LSF
QM queue (default: “local”)
Number of CPU per QM job (default: queue defaults)
Maximum amount of memory in MB to use.
Output directory (default: “./”)
Random number generator seed (default: 20170920)
show program’s version number and exit