VISTA

VISTA — versatile in-silico targeting analysis (VISTA) RNA structure-aware ML pipeline (PLS-DA) for optimized RNA sensor design.

RNA-based sensor design is often limited by the complexity of RNA structure–function relationships. VISTA combines RNA secondary structure prediction with machine learning–based feature selection (PLS-DA) to guide sensor design for optimized binding and activation. It enables high-throughput evaluation of binding sites, supports visualization of structural accessibility, and integrates ViennaRNA/NUPACK for accurate thermodynamic modeling.

This repository includes the code used for:

1. Pairwise Probability which computes pairwise probability values across an RNA target for various sensor binding sites, and maps these using an arc plot.
2. PLS-DA VISTA which runs feature selection (RFE) and a PLS-DA pipeline with logistic regression classifiers before visualizing results (loadings, latent-variable scatter, ROC curves).
3. toehold-VISTA which provides code to design toehold switches using the VISTA pipeline against new RNA targets.

---

Quickstart

Set up the toehold-VISTA environment.

Using Conda (recommended)

(requires installation of Anaconda first)

# clone the repo
git clone https://github.com/AlexGreenLab/vista.git
cd vista
# create environment
conda env create -f vista.yml
conda activate vista

Using Pip + Virtual Environment

git clone https://github.com/AlexGreenLab/vista.git
cd vista
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Using Docker

docker build -t vista:latest .
docker run --rm -it -v "$(pwd)":/app vista:latest

Other Dependencies/Packages

Installing ViennaRNA

ViennaRNA is NOT a pip package. It comes from the ViennaRNA library which provides the RNA Python module for RNA secondary structure prediction. ViennaRNA must be installed via:

Mac (Homebrew)

brew install viennarna

or via Conda:

conda install -c bioconda viennarna 

Linux (APT-based systems like Ubuntu)

sudo apt update
sudo apt install viennarna

Windows Use Windows Subsystem for Linux (WSL) and follow the Linux installation above Alternatively, download from https://www.tbi.univie.ac.at/RNA/

Installing NUPACK

Nupack is used extensively in the toehold-VISTA code. It is essential that NUPACK be downloaded. Follow instructions for download at: https://docs.nupack.org/start/#installation-requirements

Briefly, update your installation of pip and install the optional dependencies. Run the following commands:

python3 -m pip install -U pip
python3 -m pip install -U matplotlib jupyterlab

If you are using Anaconda, replace those commands with:

conda install --update-all pip matplotlib jupyterlab

Download the NUPACK package (nupack-4.0.0 or newer) into the Downloads folder, and make sure it is unzipped.

Install NUPACK 4 Python module by running the following command in your terminal:

python3 -m pip install -U nupack -f ~/Downloads/nupack-VERSION/package

Where nupack-VERSION should be replaced with the correct folder name.

Running the Code

Launching Jupyter Notebook

Jupyter Notebook provides an interactive environment to run and modify the scripts in this repository.

# Make sure your environment is active 
conda activate vista

# Install Jupyter if you don't already have it
pip install jupyter 

# Launch Jupyter Notebook
jupyter notebook

A browser window will open. Navigate to the toehold-VISTA folder to run the design code notebook. Ensure to update the "source_seq_list.csv" prior to starting the code. Follow the guides in the notebooks for running the script.