CIDMD_setup

Authors: Jesi Lee, Lee-Ping Wang

General Description

• These scripts set up CIDMD (collision-indueced dissociation molecular dynamics) simulations in TeraChem.
• Given a molecular ion (protonated molecule), CIDMD predicts a theoretical CID-MS/MS spectrum by automatically placing a collider gas argon atom and accelerating the collider towards the molecular ion using NEV ab initio MD simulations.
• To run the scripts: bash CIDMD_setup.com

Organization

For the scripts to work properly, the four files, CIDMD_setup.com, CIDMD_setup.py, CIDMD_preprun.py, and lebedev.py, should be located in the current working directory. CIDMD_setup.com contains all variables that users can set for CIDMD_setup.py to take, executes the rest of scripts, and creates number (XXX) of input files in the subfolder structures: cid/calcs/XXX/chunk_0000/ (where XXX ranges from 000 to the user-defined three digit number). To see all variables, see the Inputs section below. Each resulting subfolder contains two files: start.xyz and vel.xyz.

• start.xyz contains the coordinates for the molecular ion user-provided and the collider gas argon atom cooridiates located an user-defined distance away from the molecular ion.
• vel.xyz contains the velocity for the molecular ion user-provided (default=0) and the user-defined velocity (in AKMA unit) of the collider gas argon atom.

Inputs

• an optimized molecular ion structure in xyz file format.
• variables for CIDMD_setup.py (please see below for the variable options from the help section: python3 CIDMD_setup.py -h)

usage: CIDMD_setup.py [-h] [--extent EXTENT] [--cutoff CUTOFF]
                            [--set_distance SET_DISTANCE]
                            [--n_points N_POINTS] [--seed SEED]
                            [--deterministic] [--cartgrid CARTGRID]
                            [--lebedev LEBEDEV] [--v_scale V_SCALE] [-v]
                            xyz_in

positional arguments:
  xyz_in                REQUIRED positional argument: input .xyz structure of
                        molecular ion

optional arguments:
  -h, --help            show this help message and exit
  --extent EXTENT       Half the edge-length of cubic box in Angstrom in which
                        initial positions are generated.Should be larger than
                        the spatial extent of the molecule. (default: 50.0)
  --cutoff CUTOFF       Keep only the collider initial positions whose paths
                        come to within this distance of any atom on the
                        molecular ion (in Angstrom) (default: 2.0)
  --set_distance SET_DISTANCE
                        Atomic radius of molecular envelope surface on which
                        the collider will be placed (in Angstrom). (default:
                        5.0)
  --n_points N_POINTS   Number of initial conditions desired when using
                        randomized algorithm. (default: 50)
  --seed SEED           Number of initial conditions desired when using
                        randomized algorithm. (default: 2021)
  --deterministic       If set, generate initial positions using deterministic
                        algorithm instead of default randomized algorithm.
                        (default: False)
  --cartgrid CARTGRID   Number of Cartesian grid points for deterministic
                        algorithm. Default is set to generate 51 points as
                        -50, -48, -46, ..., 48, 50. (default: 51)
  --lebedev LEBEDEV     Lebedev grid order for deterministic algorithm (use
                        odd number in range [3,31]; higher orders not
                        recommended) (default: 5)
  --v_scale V_SCALE
  -v, --verbose         Turn on additional output for debugging. (default:
                        False)

Outputs

• a user-defined number of CIDMD input files, start.xyz and vel.xyz, varing collisional geometry.
• a user-defined number of collision trajectories that can be studied for fragmentation reaction pathways.
• further analyses can be done using this repositories: github.com/jesilee/CIDMD_analysis performs analysis of CIDMD results, and github.com/jesilee/CIDMD_compare compares the quality of CIDMD prediction against the reference mass spectra provided.

Dependencies

• python 3.6

• numpy

• these files from this repo should be in the current working directory: CIDMD_setup.com CIDMD_setup.py CIDMD_preprun.py lebedev.py

Important notes

To start CIDMD, Folder organization is very important. Here is the instruction to set up CIDMD with this repository:

1. start by creating a folder Molecule where molecule is the name of user-defined molecule. example: Psilocin

2. create a text file name mol_info.in that contains information about the molecule. example:

###__ mol_info __###
mol_name = Psilocin
mol_id   = PS
mol_mf   = C12H16N2O
mol_mw   = 204
mol_xw   = 204.268

3. git clone this current repository: git clone https://github.com/jesilee/CIDMD_setup

4. check these files are in the current working directory: CIDMD_setup.com, CIDMD_setup.py, CIDMD_preprun.py, and lebedev.py

5. decid the variable options for CIDMD (please see Inputs section) and change the variables in CIDMD_setup.com

6. execute bash CIDMD_setup.com.

7. run these trajectories with TeraChem by adding run.in file under cid/ folder. example:

coordinates start.xyz
charge      1
spinmult    1
run         md
method      ub3lyp
basis       6-31gs

# Output options
bond_order_list   yes
bond_order_thresh 0.1

# MD options #nve
timestep        1.0
velocities      vel.xyz
nstep           1000
scfintegrator   regular
removecomm      no

# SCF options
maxit           100
convthre        1e-4
# guess           ca0 cb0
purify          no
mixguess        0.0

# Technical options
# timings         yes
# gpus            2
# xtol            1e-8
precision       mixed
threspdp        1e-4
dftgrid         3

8. run this command line in cid/ folder: find calcs -name start.xyz -execdir ln -fs ../../../run.in . \; or run link_run_in.sh file from the repository.

9. execute TeraChem to run CIDMD. example: terachem run.in &> run.out

10. process CIDMD trajectories using LearnReactions.py written by Prof. Lee-Ping Wang. His repository: 'https://github.com/leeping/nanoreactor' (not yet released)

11. analyze CIDMD trajectories using these repositories: git clone https://github.com/jesilee/CIDMD_analysis git clone https://github.com/jesilee/CIDMD_compare