/BioNeMo

BioNeMo NIMs example notebooks: for optimized inference at scale

NVIDIA BioNeMo

Overview

AI is achieving incredible breakthroughs in chemistry and biology such as enabling 3D protein structure prediction, property prediction, and generation of novel protein sequences and molecules. This progress is enabling developments in the pharmaceutical industry like antibody design, small-molecule drug discovery, and newer approaches like RNA aptamer and peptide-based therapeutics. Generative models, specifically, are enabling every stage of drug discovery.

BioNeMo is a cloud service for every stage of AI-powered drug discovery. BioNeMo enables researchers with SOTA AI models for protein structure prediction, molecule generation, protein generation, and binding pose generation.

BioNeMo provides API endpoints and a rich graphical user interface (GUI) to enable browser-based access to pretrained models, and enables training and fine-tuning of these models through DGX Cloud. Users can generate molecules with MoFlow and MegaMolBART, generate protein sequences with ProtGPT-2, predict protein structures with ESMFold, OpenFold, and AlphaFold, predict properties from embeddings from MegaMolBART and ESM, and predict docked poses through DiffDock.

BioNeMo is a cloud service for every stage of AI-powered drug discovery.

Getting Started

For more information and to sign up for access to BioNeMo Service, please visit https://www.nvidia.com/en-us/gpu-cloud/bionemo/.

To get started with the BioNeMo Service API, you can explore a set of example workflows in the BioNeMo API example notebooks.