What it is
BioNeMo is NVIDIA's framework and set of NIM microservices for training and deploying biomolecular foundation models — protein LMs (ESM-2), generative chemistry (MolMIM, MegaMolBART), docking (DiffDock), structure (OpenFold) and more. It is the industrial 'picks-and-shovels' layer many drug-discovery teams build on, and a co-developer of models like Evo 2.
Evidence trail
BioAtlas keeps the path from source to decision visible. A connection records provenance; it does not imply that evidence is sufficient for every context.
Model passport
How NVIDIA BioNeMo represents biology
Category is navigation. These fields describe the model-specific computational transformation and deliberately override broad category defaults.
Biological scale
Modalities & tasks
Registry, claims and frontier intelligence
Version history not yet curated
1 version record · release year not yet normalized. Model-family identity remains separate from capability and access changes.
Explore version lineage →0 normalized claims
No task, dataset, split and metric claim has been normalized for this record yet.
Open claim intelligence →2 connected frontiers
Genome design · Peer-reviewed capability
Inspect research horizon →Connected research frontiers
These records describe active research directions, not guaranteed capabilities of this model. Evidence stages and unresolved questions are preserved separately.
Genome-scale generative biology
Arc Institute · Stanford · NVIDIA · 2026-03-01Can a foundation model read, predict and design biological sequence continuously from single nucleotides to megabase-scale genomes?
Evidence boundary and unresolved questions
Generative plausibility is not equivalent to biological viability, function or safety. Long generated sequences require extensive synthesis, containment and functional review.
- What biological constraints are learned versus memorized?
- How should whole-genome designs be evaluated before synthesis?
- Can mechanistic interpretability keep pace with model scale?
genome foundation model · long context · sequence design · biosafetyOpen frontier record →Fast, uncertainty-aware affinity screening
Terray Therapeutics research team · 2026-02-08Can virtual screening retain useful structural and affinity accuracy without expensive all-atom diffusion for every compound?
Evidence boundary and unresolved questions
The results are preprint claims; proprietary assay details and cross-lab prospective replication remain limited.
- Does coarse representation preserve water, ion, metal and covalent chemistry?
- How well does uncertainty calibrate under target and chemistry shift?
- Can active learning improve real design-make-test cycles?
virtual screening · uncertainty · active learning · coarse representationOpen frontier record →Inputs and outputs
Inputs
Model-specific biological inputsOutputs
Model inferenceTraining workflowsNIM servicesScientific and technical profile
Scientific principles
Technology
Scientific lineage
These are transparent concept matches—not claims that one scientist alone caused this model. Each connection is based on the model’s recorded domain, scientific principles, technical terms or an explicit lineage link.
Atomic structures of biologically important molecules by X-ray crystallography
Dorothy Crowfoot HodgkinStructure-based drug design depends on the experimental structural tradition she helped establish.
Intermolecular forces and excluded volume
Johannes D. van der WaalsModern force fields, docking scores, molecular dynamics and ligand–protein packing depend on these non-covalent interactions.
Lock-and-key molecular recognition
Emil FischerThe metaphor seeded structure-based ligand design, pharmacophore thinking and the search for complementary binding pockets.
Induced-fit binding
Daniel E. Koshland Jr.Flexible docking, conformational selection, protein motion and ligand-induced pocket changes are modern extensions of this idea.
Quantitative structure–activity relationships
Corwin HanschClassical QSAR established the central premise that molecular features can predict potency and guide optimization—the conceptual ancestor of modern molecular machine learning.
Denoising diffusion generative models
Jascha Sohl-Dickstein, Jonathan Ho and collaboratorsModern protein-backbone, molecular-pose and biomolecular-complex generators use diffusion to sample valid three-dimensional structures and designs.
Evaluation evidence
BioAtlas has not yet extracted a structured benchmark claim for this record.
Known limitations
- Performance depends on the evaluation dataset and operating conditions.
- A structured benchmark claim has not yet been extracted for this record.
- Outputs require task-specific scientific and experimental validation.
Milestones
Co-developed Evo 2 with the Arc Institute.
Bundles protein, molecule, docking & genomics models.