BioNeMo

BioNeMo is NVIDIA's generative AI platform for drug discovery.

This directory contains examples of running BioNeMo in a federated learning environment using NVFlare.

Notebooks

In this repo you will find two notebooks under the task_fitting and downstream folders respectively:

The task_fitting notebook example includes a notebook that shows how to obtain protein-learned representations in the form of embeddings using an ESM-2 pre-trained model.
The downstream notebook example shows three different downstream tasks for fine-tuning a BioNeMo ESM-style model.

Requirements

NOTE: This notebook is designed to run inside the BioNeMo Framework Docker container. Follow these [instructions](https://docs.nvidia.com/ai-enterprise/deployment/vmware/latest/docker.html) to set up your Docker environment and execute the following bash script before opening this notebook.

To set up your environment, simply run (outside this notebook):

./start_bionemo.sh

This script will automatically pull the BioNeMo Docker container (tested with version nvcr.io/nvidia/clara/bionemo-framework:2.5) and launch Jupyter Lab. The Jupyter Lab interface will be available at http://<your-hostname>:8888 where <your-hostname> should be replaced with your machine's hostname or IP address. Open that URL in your browser and access this notebook.

For detailed setup guidance, refer to the BioNeMo User Guide.