Deep SNUPI

Deep SNUPI is a graph neural networks model to predict the three-dimensional shape of DNA origami assemblies. It was trained by hybrid data-driven and physics-informed approach.

Prerequisite

• python=3.9
• pytorch=2.0.0
• torch-geometric==2.3.1
• roma==1.3.2
• plotly==5.15.0
• streamlit==1.24.0

Install via Conda (Recommended)

• Ensure your working-directory is in DeepSNUPI folder

• Create DeepSNUPI enviroment from environment.yml file

  conda env create -f env.yml

• Activate the environment

  conda activate DeepSNUPI

• Setup Pytorch 2.0.0 for Windows

  # CUDA 11.7
  conda install pytorch==2.0.0 torchvision==0.15.0 torchaudio==2.0.0 pytorch-cuda=11.7 -c pytorch -c nvidia
  # CPU Only
  conda install pytorch==2.0.0 torchvision==0.15.0 torchaudio==2.0.0 cpuonly -c pytorch

• Setup Pytorch-Geometric 2.3.1 for Windows

  pip install torch_geometric 
  # Dependencies CUDA 11.7
  pip install torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.0.0+cu117.html
  # Dependencies CPU Only
  pip install torch_scatter torch_sparse torch_cluster torch_spline_conv -f https://data.pyg.org/whl/torch-2.0.0+cpu.html

Hardware Requirements (Recommended)

For the best performances:

• Memory (GPU): 8GB
• Memory (RAM): 128GB

Dataset

Preprocessed data

We provide the preprocessed data in the dataset/origami/ folder

• The training set: dataset/origami/training_set
• The input samples from SNUPI: dataset/origami/snupi_input_samples/

Prepare your own DNA graph representation

You can also generate your own input by SNUPI from caDNAno design

Instruction (win)

In this section we will demonstrate how to run Deep-SNUPI

• Set up your working directory in DeepSNUPI folder from command window

• Activate DeepSNUPI enviroment, run:

  conda activate DeepSNUPI
  

• To get started using DeepSNUPI by the running the following command:

  streamlit run DeepSNUPI.py
  

• Seclect GNN predict tool for predictions

• Upload your own input files or select input samples

• Click to choose prediction methods and take the predictions in output folder

References and citing

The DGNN model and the approach behind it is described in our pre-print:

Data-driven and physics-informed prediction of DNA origami shape using graph neural network