quantifying-biological-heterogeneity-single-cell-data.png

Biomedical Foundation Models

Exploring the power of BMFM technologies to drive critical tasks in drug discovery

Overview

Learning a molecular language for protein interactions is crucial for advancing drug discovery. Foundation models, trained on diverse biomedical data like antibody-antigen interactions and small molecule-protein interactions, are transforming this field. Unlike traditional computational approaches, they widen the search scope for novel molecules and refine it to eliminate unsuitable ones, emphasizing the detailed nuances in molecular structure and dynamics.

IBM Research biomedical foundation model (BMFM) technologies leverage multi-modal data of different types, including drug-like small molecules and proteins (covering a total of more than a billion molecules), as well as single-cell RNA sequence and other biomedical data.

Our research team has a diverse range of expertise, including computational chemistry, medicinal chemistry, artificial intelligence, computational biology, physical sciences, and biomedical informatics.

Our BMFM Technologies currently cover the following three domains:

BMFM_Technologies_for_Drug_Discovery (1).png

Foundation models for Targets Discovery

Targets discovery models learn the representation of DNA, bulk RNA, single-cell RNA expression data and other cell level signaling information for the identification of novel diagnostic and therapeutic targets, allowing tasks such as cell type annotation and classification, gene perturbation prediction, disease state prediction, splice variants prediction, promoter region, and treatment response.

Foundation Models for Biologics Discovery

Biologics discovery models focus on biologic therapeutics discovery, with the goal of leveraging large-scale representations of protein sequences, structures, and dynamics for diverse downstream tasks associated with multiple biologics modalities. These models produce unified representations of biological molecular entities, integrating data such as protein sequences, protein complex structures, and protein-protein complex binding free energies into a single framework. These models can serve as the basis for diverse downstream tasks in therapeutic design, including candidate generation and assessment, across antibody, TCR, vaccine, and other modalities.

Foundation Models for Small Molecules Discovery

Small molecules models can address a wide variety of downstream predictive and generative tasks. These models are trained on multiple representations of small molecules data to learn rich low-dimensional representations of biochemical entities relevant to drug discovery, allowing tasks such as property and affinity prediction, multi-model late fusion prediction, and scaffold-based generation. Predictive models are transformer models pretrained on multiple views (i.e., modalities) of small molecule data and learn rich latent representations by maximizing mutual information across different views of molecules. Generative models learn by driving input molecules to output mutant molecules with a cognate property embedding of the mutant via diffusive denoising networks. Given a set of desired properties and a desired template molecule (3D-strcutures), a set of designer molecules (3D-strcutures) can be obtained.