Quantum RNA Folding

A Streamlit app for predicting RNA secondary structure using QAOA (Quantum Approximate Optimization Algorithm) on PennyLane.

Based on Jiang et al. (2023).

Quick Start

# Install dependencies
pixi install

# Run the app
pixi run streamlit run qrna_folding/main.py

Open http://localhost:8501 in your browser.

Usage

The app follows a 3-step workflow:

1. Preprocessing

• Enter an RNA sequence using valid bases: A, U, G, C
• Set minimum stem length (default: 3)
• Click Perform Preprocessing
• View the adjacency matrix showing potential base pairs (AU/GU = 2 H-bonds, GC = 3 H-bonds)

2. Stem Selection

• Choose number of stems to analyze
• Select method:
  • longest — prioritize stems with most base pairs
  • random — random selection
• Click Select Stems
• View network graphs of selected stems

3. QAOA Experiment

• Set QAOA layers (2–9, default: 5)
• Choose device:
  • default.qubit — standard PennyLane simulator
  • lightning.qubit — high-performance C++ simulator
• Click Run Experiment
• Monitor optimization in terminal (100 Adam steps)

Project Structure

qrna_folding/
├── main.py           # Streamlit UI
├── preprocessors.py  # RNA processing, stem detection
├── experiments.py    # QAOA circuit and optimization
├── visualisation.py  # Adjacency matrix and network plots
└── utils.py          # Hamiltonian utilities

Dependencies

Runtime dependencies managed via Pixi — see pixi.toml.

Dev dependencies defined in pyproject.toml:

• ruff — linter and formatter
• pre-commit — git hooks