Practical considerations for Molecular Dynamics: Scratch Pad

Table of Contents

Flow Charts

The Global MD Algorithm

graph TD A("Input initial conditions") --> B(fa:fa-spinner) B --> C["compute forces"] C --> D["update configuration"] D --> E[output step] E -->|repeat *nsteps* times | B

Main phases of a molecular dynamics simulation

flowchart TB %% subgraph work %% direction TB subgraph systemprep["System Preparation"] direction TB A1["check input structure"] A2["generate topology"] A3["define simulation box"] A4["add ions and solvent"] A1 --> A2 A2 --> A3 A3 --> A4 end subgraph minimization["Energy Minimization and Relaxation"] direction TB B1["Energy Minimization"] B1 --> B2["short NVT (possibly w/ position restraints)"] B2 --> B3["short NPT"] end subgraph equilibration["Equilibration"] direction TB C1["Equilibration MD"] end subgraph production["Production"] direction TB D1["Production MD and data collection"] end subgraph analysis["Analysis"] direction TB E1["Data Analysis"] end systemprep --> minimization minimization --> equilibration equilibration --> production production --> analysis %% end %% work --- intent %% subgraph intent %% direction TB %% QA1(["Is the molecule whole? Are there missing atoms or residues?"]) %% QA2(["Check whether each component as integer charge."]) %% QA4(["Neutralize system."]) %% A1 -.-> QA1 %% A2 -.-> QA2 %% A4 -.-> QA4 %% B1 -.-> QB1(["Resolve atomic clashes that lead to high forces."]) %% B2 -.-> QB2(["Let system relax to target temperature."]) %% B3 -.-> QB3(["Let system relax to target density."]) %% end

The flow of data

graph LR inp1(coordinates) ---> md{MD ENGINE} inp2(topology) ---> md inp3(md parameters) ---> md md ---> out1(log file as plain text) md ---> out2(trajectory of coordinates) md ---> out3(final coordinates) md ---> out4(checkpoint / restart file) md ---> out5(energy components, measurements, ...)

MD-parameters

Time Step

The maximum time step with which MD simulations can be expected to be stable depends on the fastest oscillations of the model.

conditions typical maximum time step
simulation without bond- or angle-constraints 0.5 fs - 1 fs
covalent bonds constrained 2 fs
using virtual interaction sites ~ 5 fs
using a coarse-grained forcefield (e.g. MARTINI) 20 - 40 fs

Section “Choosing an appropriate timestep” in Braun-2019

Cut-Offs

The cut-off scheme should match what was used for parameterization of the forcefield. It has been shown that using PME for determining long-range electrostatics in many cases improves the results, even if the forcefield has not been parameterized with PME. (FIXME: citations needed)

FIXME: citations needed

Force Fields

Different force fields have been parameterized with different philosophies and tuned to reproduce different physical properties or with emphasis on different kinds of molecules.

Generally speaking, different forcefield cannot be combined unless they are members of the same family and therefore the secondary forcefield has been specifically parameterized to be compatible with the parent-forcefield.

A Selection of Common Forcefield Families