News | March 21, 2023

NVIDIA Unveils Large Language Models And Generative AI Service To Advance Life Sciences R&D

Part of NVIDIA AI Foundations, New BioNeMo Cloud Service Accelerates Life Sciences Research, Drug Discovery and Protein Engineering; Amgen and a Dozen Startups Among Early Access Customers

NVIDIA today announced an expanded set of generative AI cloud services for customizing AI foundation models to accelerate the creation of new proteins and therapeutics, as well as research in the fields of genomics, chemistry, biology and molecular dynamics.

Part of NVIDIA AI Foundations, the new BioNeMo Cloud service offering — for both AI model training and inference — accelerates the most time-consuming and costly stages of drug discovery. It enables researchers to fine-tune generative AI applications on their own proprietary data, and to run AI model inference directly in a web browser or through new cloud application programming interfaces (APIs) that easily integrate into existing applications.

“The transformative power of generative AI holds enormous promise for the life science and pharmaceutical industries,” said Kimberly Powell, vice president of healthcare at NVIDIA. “NVIDIA’s long collaboration with pioneers in the field has led to the development of BioNeMo Cloud Service, which is already serving as an AI drug discovery laboratory. It provides pretrained models and allows customization of models with proprietary data that serve every stage of the drug-discovery pipeline, helping researchers identify the right target, design molecules and proteins, and predict their interactions in the body to develop the best drug candidate.”

Amgen Among Early Users
Amgen, one of the world’s leading biotechnology companies, is already using the service to advance its research and development efforts.

“BioNeMo is dramatically accelerating our approach to biologics discovery,” said Peter Grandsard, executive director of Biologics Therapeutic Discovery, Center for Research Acceleration by Digital Innovation at Amgen. “With it, we can pretrain large language models for molecular biology on Amgen’s proprietary data, enabling us to explore and develop therapeutic proteins for the next generation of medicine that will help patients.”

Generative AI Supercharges Drug Discovery Pipeline
BioNeMo Cloud service includes pretrained AI models to help researchers build AI pipelines for drug development. It has been adopted by drug-discovery companies including Evozyne and Insilico Medicine to support data-driven drug design for new therapeutic candidates.

Generative AI models can rapidly identify potential drug molecules — in some cases designing compounds or protein-based therapeutics from scratch. Trained on large-scale datasets of small molecules, proteins, DNA and RNA sequences, these models can predict the 3D structure of a protein and how well a molecule will dock with a target protein.

New Generative AI Models Available With BioNeMo Service Early Access
BioNeMo now has six new optimized, open-source models, in addition to its previously announced MegaMolBART generative chemistry model, ESM1nv protein language model and OpenFold protein structure prediction model. They include:

  • AlphaFold2: A deep learning model that reduces the time it takes to determine a protein’s structure from years to minutes or even seconds, just by using its amino acid sequence, developed by DeepMind and already used by over a million researchers.
  • DiffDock: To help researchers understand how a drug molecule will bind with a target protein, this model predicts the 3D orientation and docking interaction of small molecules with high accuracy and computational efficiency.
  • ESMFold: This protein structure prediction model, using Meta AI’s ESM2 protein language model, can estimate the 3D structure of a protein based on a single amino acid sequence, without requiring examples of several similar sequences.
  • ESM2: This protein language model is used for inferring machine representations of proteins which are useful for downstream tasks such as protein structure prediction, property prediction and molecular docking.
  • MoFlow: Used for molecular optimization and small molecule generation, this generative chemistry model creates molecules from scratch, coming up with diverse chemical structures for potential therapeutics.
  • ProtGPT-2: This language model generates novel protein sequences to help researchers design proteins with unique structures, properties and functions.

The BioNeMo Service makes these generative AI models easily accessible through a browser-based interface for interactive inference and protein structure visualization. And by pairing BioNeMo with the supercomputing resources in NVIDIA DGX Cloud, researchers can customize their models on a fully managed software service using NVIDIA Base Command Platform and the NVIDIA AI Enterprise software suite.

Pharma Companies, Startups Tap BioNeMo to Optimize AI Workflows
Pharmaceutical companies and drug discovery startups are using BioNeMo today and, in many cases, seeing significant results.

Amgen pretrained and fine-tuned BioNeMo’s ESM model architecture using its own proprietary data on antibodies. It was able to slash the time it takes to train five custom models for molecule screening and optimization from three months to a few weeks on DGX Cloud.

Researchers at Evozyne, a Chicago-based biotechnology company and member of the NVIDIA Inception program for cutting-edge startups, have collaborated with NVIDIA to develop a BioNeMo-based deep learning model called the Protein Transformer Variational AutoEncoder. The generative AI model, fine-tuned on Evozyne’s proprietary protein data, enables the design of synthetic variants with significantly improved performance compared to enzymes found in nature.

Insilico Medicine, a premier member of NVIDIA Inception, is using BioNeMo to accelerate the early drug discovery process, which traditionally takes more than four years and costs around $500M. Using generative AI from end to end, Insilico was able to identify a preclinical candidate drug in one-third of the time and for one-tenth of the cost. The drug is expected to soon enter phase 2 clinical trials with patients.

Source: NVIDIA