Skip to content

Latest commit

 

History

History
251 lines (191 loc) · 6.5 KB

schrodinger_tips.md

File metadata and controls

251 lines (191 loc) · 6.5 KB
Raw

Schrodinger Notes

Good practices

  • Always know you working directory. An easy way to always keep track of your working directory is by modifying some settings in the Maestro preferences (under the Maestro menu bar on Mac or File in Linux and Windows).

Maestro Settings

  • Organize your files! Always keep only ONE project per directory. In this way, each parent directory will contain one project and its associated I/O files. Note that the project and its parent directory have the same name.

Using the settings above, each time you open a project Maestro will automatically change the working directory to the project's parent directory, that is the directory that contains the project. If you keep only one project per directory, then you can easily organize your files like this:

~/Documents/Schrodinger
            |--- my_project_1
            |    |-- job_1
            |    |-- job_2
            |    |-- my_project_1.prj  (<-- the project!)
            |    |-- notes.txt
            |
            |--- my_other_project_2
                 |-- job_1
                 |-- job_2
                 |-- my_other_project_2.prj
                 |-- notes.txt

Desmond

Welcome to the dark side of Desmond.

MD Restraints:

Forcefields:

Useful references:

Command-line tools:

$SCHRODINGER/run viparr.py

Will assign a custom forcefield:

  • use -f FFNAME to use a built-in ff (listed with -h)
  • use -c to run viparr on selected CT blocks.
$SCHRODINGER/run build_constraints.py

This is required to constrain hydrogens. Default settings work fine.

$SCHRODINGER/run desmond_ff_builder

Is used to assign the OPLS forcefield. --keep_ffio_block is useful if you are mixing forcefields, use this at the end of your workflow. --skip_fepio is usually required.

$SCHRODINGER/run rebuild_cms.py

If tweaking with CTs, use this with -make_full_ct to build the final "full_system" used for visualization in maestro

System setup

Example WYSIWYG configuration system.csb file. Use this with

$SCHRODINGER/run $SCHRODINGER/utilities/system_builder.

Keywords are undocumented, but things can be discovered abusing grep or ag inside schrodinger installation directory.

{
  set_oplsaa_version_as OPLS3e

  read_solute_structure "system_setup_6MVD-membrane.mae"

  solvent_desmond_oplsaa_typer {
    input_file_name "spc.box.mae"
    run
  }
  boundary_conditions orthorhombic 85.941 85.941 170.000
  positive_ion_desmond_oplsaa_typer {
    input_file_name "Na.mae"
    run
  }
  negative_ion_desmond_oplsaa_typer {
    input_file_name "Cl.mae"
    run
  }
  neutralize
  salt_positive_ion_desmond_oplsaa_typer {
    input_file_name "Na.mae"
    run
  }
  salt_negative_ion_desmond_oplsaa_typer {
    input_file_name "Cl.mae"
    run
  }
  add_salt 0.150000

  align_principal_axis_with_xyz No
  align_principal_axis_with_diagonal No
  rezero_to_center_of_mass No
  
  solvate
  write_mae_file "system_setup_6MVD_membrane-out.mae"
}

Manually build CTs ...

Enhanced Sampling

Example blocks can be placed in the .msj inside the last simulate block or anywhere in the .cfg file:

Using the BiasingForce plugin:

https://www.deshawresearch.com/downloads/download_desmond.cgi/Desmond_Users_Guide-0.5.3.pdf#subsection.2.6.2

    # BIASINGFORCE
    atom_group = [
        { atom = "asl:a.pt CA"
        index = 1
        name = cm_moi
        }
        { atom = "asl:a.e P and r.pt POPC"
        index = 2
        name = cm_moi
        }
    ]
    backend = {
        force.term = {
            list = [ BiasingForce]
            BiasingForce = {
                type = BiasingForce
                cm_moi = [ {
                    groups = [1 2]
                    distance = 25
                    distance_coeff = 10
                } ]
                output = {
                    first = 0.0
                    interval = 100.0
                    name = "$JOBNAME.bforce"
                }
            }
        }
    }

Using enhsamp (The hard way)

Use the references described above to write an enhsamp script. This example will run a simple targeted MD for one atom.

# bias.mexp
declare_output( name     = "bias",
                first    = 0.0,
                interval = 20 );

static p0(3), v(3);

p_f = array(0, 0, 0); # target positions

gids = atomsel("asl: a.n 1");

sim_time = 100;

k = 0.75;

if time()
    then {
        ref = p0 + v * time();
        p = pos(gids[0]);
        r = norm(ref - p);
        1/2 * k * r^2;
    } else {
        store(p0, pos(gids[0]));
        store(v, (p_f - pos(gids[0])) / sim_time);
        0;
    };

Use the following to translate the script in ark syntax

$SCHRODINGER/run enhsamp.py bias.mexp > bias.ark

Then place the content of bias.ark in the .cms or .cfg files (as described above).

WARNING: enhsamp.py is bugged and you should edit the first few words of its output so that the final block looks like this:

  backend.force.term = {
    list[+]=ES
    ES={type=enhanced_sampling
    gids=[...]
    sexp=[...]
    metadynamics_accumulators=[] storage={...} name="bias" first=0.0 interval=20.0}}
}

Indentation is not relevant in any way, but make sure that closing parentheses {} are paired correctly.

Change the beginning of the ark output from:

force.term{list[+]=ES ES=force.term{list[+]=ES ES={type=enhanced_sampling ...

to:

backend.force.term = { list[+]=ES ES={type=enhanced_sampling ... }

Fragment-based ligand docking

from https://www.schrodinger.com/kb/582

  • Increase the initial number of poses per ligand kept from 5000 to 50000;
  • Widen the scoring window from 100.0 to 500.0;
  • Increase the number of minimized poses per ligand to 1000;
  • Select "Use expanded sampling".

Fragments are located in $SCHRODINGER/glide-v9.3/data/glide/