search

mlsample / hairygami_umbrella_sampling

Documentation can be found in the website below
https://mlsample.github.io/hairygami_umbrella_sampling/
0 stars 0 forks source link

readme

ipy_oxDNA

This repository contains Python code for running oxDNA umbrella sampling and large throughput simulations. This code is complementary to the article:

Sample, Matthew, Michael Matthies, and Petr Šulc. "Hairygami: Analysis of DNA Nanostructure's Conformational Change Driven by Functionalizable Overhangs." arXiv preprint arXiv:2302.09109 (2023).

Within the examples/tutorials folder exist Jupyter notebook tutorials and examples. The full documentation can be found here:(https://mlsample.github.io/ipy_oxDNA/index.html).

Contents

Introduction

oxDNA is a molecular dynamics simulation code that can be used to study the mechanical and thermodynamic properties of DNA and RNA molecules. Umbrella sampling is a highly parallelizable simulation technique that is used to calculate the free energy profiles between two particles or groups of particles. The ipy_oxDNA repository provides a Python interface for running oxDNA umbrella sampling simulations, allowing users to easily perform these simulations and analyze their results.

How to run NVIDIA Multiprocessing Service (mps)

NVIDIA MPS is a specialized service offered by NVIDIA, designed to enhance the multiprocessing capabilities of CUDA-enabled GPUs. Utilizing MPS allows for the execution of multiple simulations on a single GPU, resulting in an approximate 2.5x performance increase for specific simulation techniques such as Umbrella Sampling, Metadynamics, and Multi-Replica simulations.

In the absence of MPS, running concurrent simulations on a single GPU can lead to significant performance degradation. For comprehensive details on MPS, the official documentation is available at NVIDIA MPS Documentation.

export CUDA_MPS_PIPE_DIRECTORY=/tmp/mps-pipe_$SLURM_TASK_PID
export CUDA_MPS_LOG_DIRECTORY=/tmp/mps-log_$SLURM_TASK_PID
mkdir -p $CUDA_MPS_PIPE_DIRECTORY
mkdir -p $CUDA_MPS_LOG_DIRECTORY
nvidia-cuda-mps-control -d

Prerequisites

Before using this code, you will need to have the following installed:

Installation

To install the ipy_oxDNA code, clone the repository to your local machine:

git clone https://github.com/mlsample/ipy_oxDNA.git

Install the package using: pip install .

Install the Weight Histogram Analysis Technique (http://membrane.urmc.rochester.edu/?page_id=126)

chmod +x ./install_wham.sh

./install_wham.sh

Example Notebooks

This repository includes a number of example notebooks that demonstrate how to use the code to perform oxDNA umbrella sampling simulations. These notebooks can be used as a starting point for your own simulations.

/example/origami_tile_N_nt_overhangs_gpu

Contributing

If you would like to contribute to this project, please fork the repository and submit a pull request.

Citation

If you find this code useful in your research, please cite the accompanying academic article:

Sample, Matthew, Michael Matthies, and Petr Šulc. “Hairygami: Analysis of DNA Nanostructure’s Conformational Change Driven by Functionalizable Overhangs.” arXiv preprint arXiv:2302.09109 (2023).