Tools
Our tools help researchers understand biology and make it programmable. From frontier AI models to datasets to benchmarks, each of our powerful technologies is designed to be shared freely with the global scientific community to accelerate science, driving advances in biology and frontier medical applications. Visit our platform to use these tools and accelerate your research.
AI Models
MODELS
ESM3
A generative, multimodal model trained across protein sequence, structure, and function. ESM3 learns a shared representation at a scale not previously achieved in biological modeling. Its design interface supports prompt-based generation, enabling new sequences and structures guided by user-defined constraints.
Models
ESM Cambrian
A next-generation language model trained on protein sequences at the scale of life on Earth. ESM Cambrian (ESMC) models define a new state of the art for protein representation learning, and deliver breakthrough performance and efficiency.
Models
Cytoland
Convolutional models trained to predict cellular landmarks from label-free microscopy images. Robust to variations in imaging parameters and developmental stages, Cytoland enables segmentation and tracking of diverse cell types and eliminates the need to label nuclei and cell membranes, improving the throughput of dynamic imaging screens.
Models
VariantFormer
A new AI model that leverages individual DNA sequences to predict how genes will be expressed across different tissues. It’s the first model that directly translates personal genetic variations — including rare variants — into tissue-specific gene activity patterns.
Our integrated ecosystem to build and use AI Models
Benchmarks
Biohub Benchmarks
Interactive and programmatic tools that enable standardized evaluation and comparison of machine learning models for biological applications. These benchmarks are a step toward ensuring that large-scale AI models can be harnessed to deliver genuine biological insights.
Datasets
Multimodal Standardized Data
Access thousands of standardized, multimodal datasets from CELLxGENE, CryoET Data Portal, Organelle Box and more on one unified platform. Choose from efficient options to download the data or analyze it with interactive visualization tools.
AI Workspace
Single Cell Workflows
Use the code-free AI Workspace to upload your own single-cell data and run models with Biohub-provided inference, then use our interactive explorer to analyze the output embeddings.
View more platforms, tools, and resources
Licensing
Many of our findings and innovations are freely available to the interested public. We make other innovations available for licensing by third parties so that they may be developed in the marketplace for public benefit. The technologies in our IP portfolio fall into four broad categories — biology, chemistry, engineering, and software — and include inventions both by intramural Biohub scientists and engineers and by Biohub Investigators from our university partners. For additional details or to discuss your interest in any of our technologies, please email ip@biohub.org.