Auto_MartiniM3

What is Auto_MartiniM3?

A toolkit that enables automatic generation of Martini force field for small organic molecules up to 25 heavy atoms, now in agreement with Martini 3 Force Field parameters. For details see the publication here.

Developers

• Magdalena Szczuka (University Toulouse 3, France)
• Tristan Bereau (University of Amsterdam, Netherlands)
• Kiran Kanekal (Max Planck Institute for Polymer Research, Mainz, Germany)
• Andrew Abi-Mansour (Molecular Sciences Software Institute, Virginia Tech, Blacksburg, US)

Supervisors

• Matthieu Chavent (Centre de Biologie Intégrative (CBI)), University Toulouse 3, CNRS, France)
• Pierre Poulain (Université Paris Cité, France)
• Paulo C. T. Souza (Laboratoire de Biologie et Modélisation de la Cellule, UMR 5239, ENS Lyon, France)

Installation with conda

For enabling automatic mapping with Auto-MartiniM3, you need to clone this repository and create a conda environment.

git clone https://github.com/Martini-Force-Field-Initiative/Automartini_M3.git
cd Automartini_M3
conda env create -f environment.yaml

This will create a conda environment called automartiniM3 which you can activate with

conda activate automartiniM3

Testing

To run the test cases and validate your installation, you will need to have pytest installed. If you installed auto_martiniM3 with conda, then pytest should already be available in your environment.

To initiate the testing, run the following:

pytest -v tests

All tests should pass within few minutes. If any of the tests fail, please open an issue.

Command-line Interface

You can invoke auto_martiniM3 from the command-line via:

python -m auto_martiniM3 [mode] [options]

By default, mode is set to 'run', which computes the MARTINI 3 force field for a given molecule.

To display the usage-information (help), either supply -h, --help, or nothing to auto_martiniM3:

usage: auto_martiniM3 [-h] [--mode {run}] [--sdf SDF | --smi SMI] [--logp LOGP] [--mol MOLNAME] [--aa AA] [-v]
                      [--fpred] [--bartender] [--simple] [--canon]

Generates Martini 3 force field for atomistic structures of small organic molecules

optional arguments:
  -h, --help     show this help message and exit
  --mode {run}   mode: run (compute FF)
  --sdf SDF      SDF file of atomistic coordinates
  --smi SMI      SMILES string of atomistic structure
  --logp LOGP    File with partial smiles and associated logP
  --mol MOLNAME  Name of CG molecule
  --aa AA        filename of all-atom structure .gro file
  -v, --verbose  increase verbosity
  --fpred        Atomic partitioning prediction
  --bartender    Bartender input file
  --simple       Simple model without dihedrals nor virtual sites
  --canon        Translate to RdKit canon structure

Developers:
===========
Magdalena Szczuka (magdalena.szczuka [at] univ-tlse3.fr)
Tristan Bereau (bereau [at] mpip-mainz.mpg.de)
Kiran Kanekal (kanekal [at] mpip-mainz.mpg.de)
Andrew Abi-Mansour (andrew.gaam [at] gmail.com)

Example

To coarse-grain a molecule, simply provide its SMILES code (option --smi SMI) or a .SDF file (option '--sdf file.sdf). You also need to provide a name for the CG molecule (not longer than 5 characters) using the --mol option. For instance, to coarse grain aspirin, you can either obtain/generate (e.g., from Open Babel) an SDF file:

python -m auto_martiniM3 --sdf aspirin.sdf --mol ASP 

(the name ASP is arbitrary) or use its SMILES code within double quotes

python -m auto_martiniM3 --smi "CC(=O)OC1=CC=CC=C1C(=O)O" --mol ASP 

In case no problem arises, it will output the gromacs ASP.itp file:

; GENERATED WITH Auto_Martini M3FF for ASP
; Developed by: Kiran Kanekal, Tristan Bereau, and Andrew Abi-Mansour
; updated to Martini 3 force field by Magdalena Szczuka
; supervised by Matthieu Chavent, Pierre Poulain and Paulo C. T. Souza 
; SMILES code : CC(=O)OC1=CC=CC=C1C(=O)O


[moleculetype]
; molname       nrexcl
  ASP          1

[atoms]
; id      type   resnr residue atom    cgnr    charge  mass ;  smiles    ; atom_num
   1       SN5a    1   ASP     N01       1        0    54   ;   CC=O     ; atoms: C0, C1, O2,          
   2       TP2a    1   ASP     P01       2        0    36   ;   CO       ; atoms: O3, C4,          
   3       TC5     1   ASP     C01       3        0    36   ;   C=C      ; atoms: C5, C6,          
   4       SC5     1   ASP     C02       4        0    54   ;   CC=C     ; atoms: C7, C8, C9,          
   5       SN6d    1   ASP     N02       5        0    54   ;   O=CO     ; atoms: C10, O11, O12, ; ALOGPS defined bead


[bonds]
;  i   j     funct   length   force.c.
   1   2     1       0.27       25000.00
   2   3     1       0.27       25000.00
   2   4     1       0.33       10000.00
   3   4     1       0.28       100000.00
   4   5     1       0.35       5000.00
#ifndef FLEXIBLE
[constraints]
#endif
;  i   j     funct   length

[angles]
;  i  j  k    funct  angle  force.c.
   1  2  5       1    95.8   100.0
   1  4  5       1    52.7    25.0

[dihedrals]
;  i  j  k  l  funct  angle  force.c.
   1  2  3  4    2    -135.5   25.0   
   2  3  4  5    2    -0.5     25.0   

[exclusions]
  1 4

The code will also output a corresponding .gro file for the coarse-grained coordinates. Atomistic coordinates can be written using the --aa output.gro option.

If --bartender flag is used, additional file for further optimization of bonded parameters with Bartender Pereira et al., 2024 will be produced. You will find more information about Bartender in the official tutorial. Examplary Bartender input file, created by Auto-MartiniM3 for aspirin and saved as ASP_bartender.inp :

# INPUT data for bonded parameter definition by BARTENDER for molecule ASP
BEADS
1 1,2,3,14,15,16
2 4,5
3 6,7,17,18
4 8,9,10,19,20
5 11,12,13,21
BONDS
1,2
2,3
2,4
3,4
4,5
ANGLES
1,2,5
1,4,5
IMPROPERS
1,2,3,4
2,3,4,5

Caveats

For frequently encountered problems, see FEP.