FastMDXplora

Fully Automated SysTem for Molecular Dynamics eXploration

Give it a structure — or just a PDB identifier — and it prepares the system, runs the dynamics, measures the trajectory, and writes the study up. The parts that usually need an expert are done for you, and refused where the structure does not say enough to do them.

The quickest way in

conda create -n fastmdxplora -c conda-forge fastmdxplora
conda activate fastmdxplora
fastmdx gui

A tab opens. Type 1L2Y, press Run, and watch a protein fold. Everything FastMDXplora does is on that page: designing a run, starting it, watching the molecule move, reading the results.

Not a dashboard attached to a command-line tool. The form is generated from the same schema the CLI validates against, so every system FastMDXplora can study can be built in the browser – a protein, a protein and its ligand, a membrane protein, an umbrella or metadynamics or steered study, a trajectory from another engine, a sweep across many systems – and every setting is reachable. What it hands back is a config file, and fastmdx explore --config runs those same bytes anywhere. Decide the study where it is comfortable to think; run it where the compute is.

If you would rather use a terminal:

fastmdx explore --system 1L2Y --output runs/trpcage

Your first run takes either route from nothing to a finished report in about ten minutes.

The config is the study

Both routes above are building the same thing. A FastMDXplora config describes a study completely – the system, its preparation, the simulation, what is measured, how it is reported – and the four phases run themselves from it. The GUI, the command line and the Python API each build one and each run all four phases; a config can also be written by hand, since the YAML is short. Every run writes resolved_config.yml with every setting filled in, so a study can be repeated from what it left behind rather than from what somebody remembers typing.

The GUI is worth using even for a command-line or Python workflow: it is the one place where every setting is visible, explained, and checked before you leave the page. See Configuration.

Finding your way around

• New here? Installation, then Your first run.

• Want to see what it can do? The FastMDXplora GUI and The four phases. Every step explains itself as it runs, so Your first run is also the shortest way to learn what the steps are for.

• Running something real? Beyond a box of water for restraints, membranes, and enhanced sampling — umbrella, steered and metadynamics; Production and GPUs for long runs.

• Reading results? Reading the results for what a run leaves behind and what makes a number one; Protein-ligand interactions for the measure that carries the most criteria.

• Looking for a flag or a setting? CLI reference, Configuration, or fastmdx explore --help, which is generated and therefore never out of date.

Start here

• Installation
  • Recommended: conda-forge
  • Installing from PyPI
  • Working on FastMDXplora itself
  • Platforms
  • Windows
  • Checking what you have
  • If something goes wrong
  • Where to go next
• Your first run
  • Install
  • The quickest route: the GUI
  • The same thing from the command line
  • What you get
  • It says why, while it happens
  • A protein with a ligand
  • Doing it repeatedly
  • When something goes wrong
  • Where next
• The four phases
  • setup
  • simulation
  • analysis
  • report
  • What a run leaves behind

Running simulations

• Simulations beyond a box of water
  • Restraints
  • Membrane systems
  • Metadynamics
  • Pulling along a coordinate
  • Umbrella sampling
  • When a run fails
• Production runs and GPUs
  • Where a production run belongs
  • Before committing hours to it
  • Launching one
  • Watching it
  • Choosing a length
  • When it stops early
  • Several runs at once
  • See also

Reading the results

• Reading the results
  • What a run leaves behind
  • Open these first
  • How a number tells you what it is worth
  • Where each kind of result is explained
  • Comparing runs
• Protein-ligand interactions
  • What it measures
  • The chemistry comes first
  • Some measurements are refused
  • Occupancy carries its observation
  • Binding modes and transitions
  • What a run looks like
  • Settings
  • Why not PLIP or ProLIF

The Config

• The Config
  • The shape of it
  • Where the settings come from
  • Checking a config without running it
  • Per-analysis settings
  • Several systems
  • Reproducing a run
• Worked examples
  • A protein on its own
  • A protein with a ligand
  • A membrane protein
  • Holding a structure still while it settles
  • Metadynamics
  • Umbrella sampling
  • Steered molecular dynamics
  • Analysing a trajectory you already have
  • Choosing what gets measured
  • Several systems at once
  • Comparing runs
  • Reproducing a run exactly
  • From Python

Interfaces

• The FastMDXplora GUI
  • Designing a run
  • Watching it happen
  • Reading the results
  • Where it runs, and who can reach it
  • When there is nothing to show
  • The API behind it
• Command-line reference
  • The commands
  • How the flags are named
  • Input and output
  • Running part of it
  • A first run that finishes quickly
  • The flags you will actually use
  • Checking things
  • See also
• Python API
  • Constructing it
  • Running part of it
  • What comes back
  • Watching a run from Python
  • Reference
  • See also

Running it elsewhere

• Running somewhere else
  • The shape of it
  • Installing where there is no network
  • Structures, offline
  • Making sure the GPU is being used
  • Batch schedulers
  • Watching from somewhere else
  • What a run leaves behind

For maintainers

• Implementation note: protein-ligand interactions
  • What the analysis answers
  • Why FastMDXplora implements this rather than depending on PLIP or ProLIF
  • What FastMDXplora knows that a general tool has to infer
  • When the trajectory came from somewhere else
  • The criteria, and where they come from
  • Geometry worth stating
  • Occupancy carries its observation
  • What checking against the other tools found
  • References