Models for Nvidia BioNeMo
Optimised AI models for healthcare are now easier to use than ever thanks to Amazon Web Services' integration of NVIDIA NIM, a collection of cloud-native microservices.NIM, a part of the NVIDIA AI Enterprise software platform available on the AWS Marketplace, provides developers with access to a growing library of AI models via industry-standard application programming interfaces, or APIs. The library provides foundation models for drug development, medical imaging, and genomics with enterprise-grade security and support.
NIM may now be accessed via Amazon SageMaker, a fully managed service for preparing data and creating, training, and deploying machine learning models, as well as AWS ParallelCluster, an open-source platform for managing and delivering high performance computing clusters on AWS. AWS HealthOmics, a service created especially for biological data processing, is another solution for coordinating NIMs.
Thanks to simple access to NIM, the hundreds of healthcare and life sciences companies that already utilise AWS will be able to use generative AI more quickly, doing away with the headaches of model construction and production packaging. It will also let developers create processes that combine data from many modalities, such as amino acid sequences, MRI scans, and plain-text patient health records, with AI models.
This effort broadens the selection of NVIDIA Clara accelerated healthcare applications and services that are offered on AWS. It was introduced today at the AWS Life Sciences Leader Symposium in Boston. These services include NVIDIA MONAI for medical imaging workflows, NVIDIA Parabricks for faster genomics, and NVIDIA BioNeMo's quick and easy-to-deploy NIMs for drug development.
NVIDIA AI Being Used by Biotech and Pharmaceutical Companies on Amazon
Nvidia A generative AI platform called BioNeMo facilitates the optimisation and training of biology and chemistry models using private data. It is made up of efficient training recipes, domain-specific data loaders, training frameworks, and foundation models. It is used by more than a hundred organisations globally.
Using the Nvidia BioNeMo framework, Amgen, one of the leading biotechnology companies globally, has trained generative models for protein design and is exploring the prospect of combining Nvidia BioNeMo with AWS.
The pretrained and optimised Nvidia BioNeMo models for molecular docking, generative chemistry, and protein structure prediction can be used on any NVIDIA GPU or GPU cluster. You can access them as NIM microservices. By combining these models, a thorough strategy for AI-accelerated drug discovery may be made possible.
A-Alpha Bio is a biotechnology company that quantifies, predicts, and designs protein-to-protein interactions using synthetic biology and artificial intelligence (AI). When researchers moved from a generic version of the ESM-2 protein language model to one that was optimised by NVIDIA and operated on NVIDIA H100 Tensor Core GPUs on AWS, they saw a speedup of more than 10x. Consequently, the group is able to try a much greater variety of protein options than they might have otherwise.
With the use of retrieval-augmented generation, or RAG also known as a lab-in-the-loop architecture developers can enhance a model for use by organisations who want to add their own experimental data to these models.
Using the Nvidia BioNeMo framework, Amgen, one of the leading biotechnology companies globally, has trained generative models for protein design and is exploring the prospect of combining Nvidia BioNeMo with AWS.
The pretrained and optimised Nvidia BioNeMo models for molecular docking, generative chemistry, and protein structure prediction can be used on any NVIDIA GPU or GPU cluster. You can access them as NIM microservices. By combining these models, a thorough strategy for AI-accelerated drug discovery may be made possible.
A-Alpha Bio is a biotechnology company that quantifies, predicts, and designs protein-to-protein interactions using synthetic biology and artificial intelligence (AI). When researchers moved from a generic version of the ESM-2 protein language model to one that was optimised by NVIDIA and operated on NVIDIA H100 Tensor Core GPUs on AWS, they saw a speedup of more than 10x. Consequently, the group is able to try a much greater variety of protein options than they might have otherwise.
With the use of retrieval-augmented generation, or RAG also known as a lab-in-the-loop architecture developers can enhance a model for use by organisations who want to add their own experimental data to these models.
Parabricks Enables Accelerated Genomics Pipelines
NVIDIA NIM comes with NVIDIA Parabricks genomics models, which are available on AWS HealthOmics as Ready2Run workflows that enable customers to set up pre-made pipelines.
The life sciences company Agilent used Parabricks genomics analysis tools running on NVIDIA GPU-powered Amazon Elastic Compute Cloud (EC2) instances to significantly boost the processing rates for variant calling workflows on its cloud-native Alissa Reporter software. When Parabricks and Alissa secondary analysis methods are integrated, researchers may quickly analyse data in a secure cloud environment.
AI-powered digital assistants can enhance healthcare by answering patient questions and helping physicians with administrative tasks. Following their training on RAG-specific data from healthcare companies, they were able to connect to relevant internal data sources to compile findings, identify trends, and increase productivity.
Creative AI startup Hippocratic AI is now testing AI-powered healthcare agents that focus on various tasks like preoperative outreach, post-discharge follow-up, and wellness coaching.
The generative AI agent for digital health is being powered by the company's implementation of NVIDIA ACE microservices and Nvidia BioNeMo Models. The business uses GPUs from NVIDIA via AWS.
The life sciences company Agilent used Parabricks genomics analysis tools running on NVIDIA GPU-powered Amazon Elastic Compute Cloud (EC2) instances to significantly boost the processing rates for variant calling workflows on its cloud-native Alissa Reporter software. When Parabricks and Alissa secondary analysis methods are integrated, researchers may quickly analyse data in a secure cloud environment.
Conversational AI in Artificial Intelligence Advances Digital Health
NIM microservices offer models that can scan genetic sequences and proteins, as well as optimised large language models for conversational AI and visual generative AI models for avatars and digital humans.AI-powered digital assistants can enhance healthcare by answering patient questions and helping physicians with administrative tasks. Following their training on RAG-specific data from healthcare companies, they were able to connect to relevant internal data sources to compile findings, identify trends, and increase productivity.
Creative AI startup Hippocratic AI is now testing AI-powered healthcare agents that focus on various tasks like preoperative outreach, post-discharge follow-up, and wellness coaching.
The generative AI agent for digital health is being powered by the company's implementation of NVIDIA ACE microservices and Nvidia BioNeMo Models. The business uses GPUs from NVIDIA via AWS.
The group used NVIDIA Audio2Face facial animation technology, NVIDIA Riva automated voice recognition, text-to-speech capabilities, and other features to enable the conversation about an avatar healthcare assistant.
NVIDIA developed a suite of resources known as Nvidia BioNeMo models specifically for application in drug development and other life sciences research. They are built on top of NVIDIA's Nemo Megatron framework, a toolset for developing and optimising large-scale language models.
Nvidia BioNeMo Models' Features
AI models that are preconditioned
These models have previously been trained on vast amounts of biological data. Subsequently, these models can be used for a variety of tasks, including as predicting protein function, evaluating the effects of mutations, and identifying potential therapeutic targets. Among the Nvidia BioNeMo models that have been pre-trained are:DNABERT: This model is helpful in understanding and predicting how DNA sequences work.
ScBERT: Because it was created using data from single-cell RNA sequencing, this model can be used to distinguish between different cell types and predict the effects of gene knockouts.
EquiDock: This method can be used to estimate the three-dimensional structure of protein interactions, which is helpful in identifying potential treatment alternatives.
The BioNeMo Service
Using the cloud-based BioNeMo Service, researchers may easily access and employ Nvidia BioNeMo's pre-trained models through a web interface. This service can be especially useful for researchers who lack access to the computational capacity required to train their own models.Taking everything into account, Nvidia BioNeMo models are a useful tool that may be used to accelerate drug discovery research. These models facilitate faster and more efficient discovery of new therapeutic targets by researchers.
News Source : Nvidia BioNeMo
0 Comments