Artificial intelligence is transforming drug discovery and life sciences, helping researchers better understand how proteins interact with other biological molecules to accelerate therapeutic development.
One of the latest advances is OpenFold3 – an open-source co-folding model that predicts the three-dimensional structure of proteins bound to other molecules, including DNA, RNA, antibodies, proteins, and small-molecule ligands. These predictions help researchers study complex biological interactions that are fundamental to disease research, drug discovery, and precision medicine.
To simplify access to this technology, AMD AIM for OpenFold3 is now available as a pre-configured application through the Vultr Kubernetes Engine (VKE) Marketplace. This builds on our recent announcement bringing AMD Enterprise AI software to the VKE Marketplace, further expanding the AMD AI ecosystem available on Vultr.
Bringing OpenFold3 to Vultr Kubernetes Engine
The latest addition to AMD's expanding AIM ecosystem, AMD AIM for OpenFold3 is now available through the Vultr Marketplace — giving researchers a seamless way to deploy it onto an existing VKE cluster with a dedicated UI for an intuitive user experience.
The application leverages AMD Inference Microservices (AIMs) to serve the OpenFold3 model on AMD Instinct™ GPUs, providing a browser-based interface for submitting protein structure prediction jobs.
Instead of manually deploying Kubernetes resources, configuring networking, or setting up inference services, researchers can deploy a pre-configured application and begin running protein structure prediction workloads.
Built for healthcare and life sciences
Unlike earlier protein structure prediction models that focused primarily on individual proteins, OpenFold3 is designed to predict the structures of biomolecular complexes. It models interactions between proteins and other biological molecules, including DNA, RNA, antibodies, and small-molecule ligands, providing researchers with a more complete view of biological systems.
These predictions support a wide range of healthcare and life sciences applications, from therapeutic discovery and antibody research to protein engineering and the study of disease mechanisms. By modeling complex molecular interactions, researchers can gain deeper insights into the biological processes that underpin disease and the development of new therapies.
Simplified deployment with AMD OpenFold3 and AIMs
OpenFold3, powered by AMD Inference Microservices (AIMs) and available through the Vultr Kubernetes Engine (VKE) Marketplace, simplifies the deployment of protein structure prediction workloads.
After selecting the application, the deployment provisions the required Kubernetes resources, configures secure HTTPS access, deploys the OpenFold3 inference service using AMD Inference Microservices (AIMs), and provides a browser-based interface for submitting prediction jobs.
Researchers can submit protein, RNA, DNA, or ligand sequences and choose from multiple sequence alignment (MSA) options depending on their workflow.
From infrastructure to discovery
Running advanced AI models shouldn't require researchers to spend time assembling infrastructure.
By combining AMD AIMs with Vultr Kubernetes Engine, organizations can deploy OpenFold3 on production-ready Kubernetes infrastructure using AMD Instinct™ GPUs through a streamlined Marketplace experience.
This enables research teams to focus on protein structure prediction and biomolecular research while relying on a pre-configured platform for deployment and management.
Ready to get started?
Deploy the AMD AIM for OpenFold3 directly from the Vultr Kubernetes Engine (VKE) Marketplace to quickly provision a production-ready environment for protein structure prediction on AMD Instinct™ GPUs. Learn how to deploy the application in the Vultr OpenFold3 deployment guide.