just alvin
November 18, 2024 22:17
NVIDIA unveils the open source BioNeMo framework, which aims to transform the biopharmaceutical industry by accelerating drug discovery and molecular design through advanced AI and supercomputing.
NVIDIA announced the launch of the open source BioNeMo framework, a groundbreaking initiative that will transform the global biopharmaceutical and scientific industries. According to the NVIDIA Newsroom, this development will accelerate drug discovery and molecular design through the integration of advanced AI and supercomputing technologies.
Industry adoption and contribution
The BioNeMo framework is already gaining traction among leading pharmaceutical and technical bio industry players, including Argonne National Laboratory, Flagship Pioneering, and Weights & Biases. Together with academic pioneers and AI researchers, these organizations are leveraging the framework to advance biomolecular science and drive innovation in drug discovery.
Argonne National Laboratory contributed a billion-parameter biological model that requires specialized software to train across high-performance computing environments. Arvind Ramanathan, leader of Argonne’s computational science group, emphasized the importance of BioNeMo in enabling the biotechnology community to scale the training of large-scale biological models without extensive computational expertise.
BioNeMo platform and its features
The comprehensive BioNeMo platform is designed to enhance AI model creation, customization, and deployment for drug discovery and molecular design. By integrating with accelerated computing infrastructure, the platform significantly reduces cost, increases scale, and accelerates drug discovery workflows to deliver trusted insights from biomolecular data.
In addition to the BioNeMo framework, the platform integrates NVIDIA NIM™ microservices optimized for secure, scalable AI inference and the NVIDIA BioNeMo Blueprint, which provides reference designs for wet lab and compute workflows.
Advanced AI models and microservices
NVIDIA has introduced a suite of optimized NIM microservices that facilitate rapid deployment across a variety of environments, whether on-premises or in the cloud. This flexibility accelerates the time from inference to insight, which is important for biological drug discovery studies.
Among the supported models is AlphaFold2, developed by Google DeepMind, which improves protein structure prediction. Based on MIT research, the DiffDock 2.0 model improves molecular orientation predictions, while RFdiffusion and ProteinMPNN microservices support the design of new proteins for therapeutics.
Extensive integration and support
More than 200 tech bio companies, large pharmaceutical companies, and startups are integrating BioNeMo into their drug discovery platforms. Global systems integrators and cloud service providers, including Accenture, AWS, and Deloitte, are accelerating the deployment of NVIDIA BioNeMo Blueprints to enterprises around the world, further expanding their impact across the industry.
Image source: Shutterstock