Multimodalities Models
Multimodal biological foundation models integrate information across molecular layers — including DNA sequence, gene expression, chromatin accessibility, and protein structure — to build richer representations of biological systems than any single data type affords. This cross-modal learning enables tasks that span scales, such as predicting how genetic variants propagate through gene regulatory networks to alter cellular phenotypes. As multi-omics datasets become routine, these models are emerging as essential tools for systems-level understanding of health and disease.
12 models in this category
Notable Models
Top-rated multimodalities models from our evaluations
NatureLM
Microsoft Research AI for Science
Unified science foundation model from Microsoft Research treating molecules, proteins, RNA, DNA, and materials as a shared sequence language for cross-domain generation.
rBio
Chan Zuckerberg Initiative
A reasoning language model post-trained on virtual cell simulations to answer complex biological questions about gene perturbations in natural language.
BioMed Multi-View
IBM Research
Multi-view molecular foundation model that integrates graph, image, and text representations via late fusion for molecular property and target prediction.
BioT5+
Microsoft Research Asia
An enhanced T5-based encoder-decoder that unifies molecule, protein, and text understanding via IUPAC integration and multi-task instruction tuning.
ChatCell
ZJUNlp
A T5-based conversational framework that converts scRNA-seq data into cell sentences, enabling cell type annotation, pseudo-cell generation, and drug sensitivity prediction via natural language.
MAMMAL
IBM Research
Multi-modal, multi-task biological foundation model trained on 2 billion samples spanning proteins, small molecules, and single-cell gene expression.