GRANA

Graphical Recognition and Analysis of Nanostructural Assemblies

GRANA is an AI-enhanced, user-friendly software tool designed to recognize and analyze grana stacks on electron micrographs, generating a detailed set of structural measurements

This software is described in the reference paper titled "GRANA: Accelerating Chloroplast Grana Nanomorphology Analysis with Hybrid Intelligence," currently under review for publication.

How to Get Started

Demo Images Dataset

Try GRANA with a set of Demo TEM Images available for download here. The scale for the images is 1.298 pixels per nanometer

Demo

Explore the GRANA tool’s demo version on Hugging Face Spaces, allowing you to analyze up to five images simultaneously

Docker Container (recommended)

The recommended way to run GRANA is through a Docker container, allowing for the analysis of an unlimited number of images

GitHub Repository

Visit the GitHub repository for the latest releases, source code, and further details on installation and usage

Running GRANA on your local machine

We strongly recommend installing locally with Docker to get the best experience, as the Huggingface demo runs significantly slower. Running GRANA on your machine ensures smoother and faster performance.

Installing via Docker Container

The recommended method for running GRANA is through a Docker container. If Docker is already installed on your system, simply enter the following command in your Terminal:

docker run -p 7860:7860 mbuk/grana

Once the command is executed, you can access the GRANA interface in your browser at http://localhost:7860.

If you’re new to Docker

Refer to the Docker Desktop official documentation for detailed installation steps based on your operating system.

Additional Resources

  • Documentation: The GRANA GitHub repository contains the source code and documentation.
  • Demo Dataset: Try GRANA with a set of Demo Images available here. We encourage you to test GRANA with your own TEM images. Detailed guidance on best practices for microscopy imaging to ensure efficient and reliable analysis with the GRANA tool can be found in the reference paper and Supplementary Figure 2. For best results, use well-contrasted TEM micrographs with clearly visible grana structures. Avoid images showing entire chloroplasts, as grana may appear too small for accurate structural analysis due to limited layer and lumen visibility.

For additional support, refer to the README on the GitHub repository, and follow the documentation links for troubleshooting and usage tips.

Key Features of GRANA

AI-Powered Analysis

GRANA utilizes three artificial neural networks, each optimized for a specific stage in grana identification and measurement, delivering fast and reliable results for large datasets

User-Friendly Interface

Designed to simplify the complex process of grana measurement and analysis into a seamless, one-click workflow

Scalable

With GRANA, analyze grana thylakoids from various species and environmental conditions over 100 times faster than traditional manual approaches. Deploy GRANA on your local machine or any environment that supports Docker