pyVLMP 1.0.59

Virtual Laboratory for the simulation of Macromolecular Properties

VLMP (Virtual Laboratory Massively Parallelized)

Table of Contents

• Introduction
• Installation
• Getting Started
• Workflow
• License
• Contact

Introduction

VLMP is a Python library designed for running parallelized simulations, specifically optimized for molecular dynamics and other continuous models. Built on the backend technology of UAMMD-structured, it leverages multi-level parallelization to achieve highly efficient simulation runs.

Features

• Multi-level Parallelization: Run multiple simulations concurrently on a single GPU or distribute across multiple GPUs.
• Optimized for Coarse-grained Models: Achieve better GPU utilization with small-scale simulations.
• Highly Configurable: Easily adaptable for a variety of scientific phenomena.
• Community Sharing: Distribute new models as VLMP modules.

Documentation

Coming soon.

Installation

Prerequisites

Install UAMMD-structured before proceeding. UAMMD-structured Documentation

Installing VLMP

Via pip:

pip install pyVLMP

Or clone the GitHub repository:

git clone https://github.com/PabloIbannez/VLMP.git
cd VLMP
pip install .

Verifying Installation

import VLMP

Getting Started

Here's a minimal example to simulate a set of DNA chains:

   import VLMP
   from VLMP.utils.units import picosecond2KcalMol_A_time
   from numpy import random

   # Convert picoseconds to AKMA time unit
   ps2AKMA = picosecond2KcalMol_A_time()

   # Number of sequences and sequence set size
   Nsequence = 10
   sequenceSetSize = 10

   # Length of each sequence and the basis of DNA
   sequenceLength  = 100
   basis = ['A', 'C', 'G', 'T']

   # Generate random sequences
   sequences = []
   for i in range(Nsequence):
       sequences.append(''.join(random.choice(basis, sequenceLength)))

   # Populate simulation pool
   simulationPool = []
   for seq in sequences:
       # Configure simulation parameters
       simulationPool.append({
           "system": [
               {"type": "simulationName", "parameters": {"simulationName": seq}},
               {"type": "backup", "parameters": {"backupIntervalStep": 100000}}
           ],
           "units": [{"type": "KcalMol_A"}],
           "types": [{"type": "basic"}],
           "ensemble": [
               {"type": "NVT", "parameters": {"box": [2000.0, 2000.0, 2000.0],
                                              "temperature": 300.0}}
           ],
           "integrators": [
               {"type": "BBK", "parameters": {"timeStep": 0.02*ps2AKMA,
                                              "frictionConstant": 0.2/ps2AKMA,
                                              "integrationSteps": 1000000}}
           ],
           "models": [
               {"type": "MADna", "parameters": {"sequence": seq}}
           ],
           "simulationSteps": [
               {"type": "saveState", "parameters": {"intervalStep": 10000,
                                                    "outputFilePath": "traj",
                                                    "outputFormat": "dcd"}},
               {"type": "thermodynamicMeasurement", "parameters": {"intervalStep": 10000,
                                                                   "outputFilePath": "thermo.dat"}},
               {"type": "info", "parameters": {"intervalStep": 10000}}
           ]
       })

   # Initialize VLMP and load simulation pool
   vlmp = VLMP.VLMP()
   vlmp.loadSimulationPool(simulationPool)

   # Distribute simulations and set up
   vlmp.distributeSimulationPool("size", sequenceSetSize)
   vlmp.setUpSimulation("EXAMPLE")

Execute the simulations with:

cd EXAMPLE
python -m VLMP -s VLMPsession.json --local --gpu 0 1

Workflow

1. Simulation Configuration: Define simulation parameters.
2. Simulation Pool Creation: Prepare multiple configurations for batch execution.
3. Simulation Distribution: Distribute simulations across computational resources.
4. Simulation Execution: Execute simulations on GPU using UAMMD-structured.

License

GPLv3

Contact

For issues and contributions, please contact: GitHub Issues