Skip to main content
BIO·ATLASby Bayes Pharma.ai
Search
AccountStart evaluation
AI drug discovery models/Protein Foundation & Representation/ESM-2
model-family passport · Review date not recorded

ESM-2

A foundational protein language model for residue- and sequence-level representations.

4/7Evidence fields documented
60-SECOND EVALUATION VIEW

What should a scientist know before using ESM-2?

SupportedEvidence supports the stated context with explicit boundaries
Best suited forRepresentation · Prediction
Evidence supportsProtein representation and structure evaluations: Peer-reviewed
Evidence does not establishUniversal superiority, therapeutic success, clinical utility or regulatory acceptance.
Major limitationPerformance depends on the evaluation dataset and operating conditions.
CompareAdd to evaluationView evidence
Current registry recordVersion history not yet curated1 recorded release · Review date not recorded. A newer version is not assumed to be universally better.

What it is

ESM-2 learns contextual protein representations from amino-acid sequences with masked-language modelling.

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.

Sources3 connectedPrimary resources and normalized claims
Claims1 normalizedIntegrated discovery platform
EntityESM-2model-family · Version history not yet curated
ReviewReview date not recordedReview date not claimed
ConclusionContext requiredAdd to an evaluation before operational use

Model passport

Entity typemodel-family
OrganizationMeta AI (FAIR)
Model family introducedNot normalized
AccessOpen source
Commercial useAllowed / verify checkpoint terms
DeploymentSelf-hosted
ComputeGPU recommended
Domainsprotein
Biology → representation → computation → evidence

How ESM-2 represents biology

model-familyprotein

Category is navigation. These fields describe the model-specific computational transformation and deliberately override broad category defaults.

1 · Biological inputs
Amino-acid sequence
2 · Input representation
Amino-acid tokens
3 · Internal representation
Contextual protein embeddings
4 · Architecture
Protein Transformer
5 · Learning objective
Self-supervised protein language modelling
6 · Output representation
Dense vectorsScores

Biological scale

proteinresidue

Modalities & tasks

ProteinRepresentationPrediction

Registry, claims and frontier intelligence

Versioned registry

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 →
Benchmark claim ledger

1 normalized claim

Integrated discovery platform · Protein representation and structure evaluations

Open claim intelligence →

Inputs and outputs

Inputs

Amino-acid sequence

Outputs

Protein embeddingsTask predictions

Scientific and technical profile

Scientific principles

Protein language modellingSelf-supervised learning

Technology

Transformer encoderMasked language modelling
Ideas before algorithms

Scientific lineage

Explore all foundations

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.

Computational intelligence

Information, entropy and communication

Claude E. Shannon

Sequence modelling, cross-entropy training, language models, mutual information and representation learning all use Shannon’s framework.

Matched concepts: language model, sequence, representation
Structural biology

Anfinsen’s dogma—the thermodynamic hypothesis

Christian B. Anfinsen

Protein structure prediction, inverse folding and generative protein design all assume that sequence strongly constrains structure and function.

Matched concepts: sequence, fold, protein
Computational intelligence

Transformer self-attention

Ashish Vaswani and colleagues

Protein, genome, molecule and single-cell foundation models use attention to learn dependencies across biological sequences and multimodal inputs.

Matched concepts: transformer, language model, sequence

Evaluation evidence

Dataset or evaluationProtein representation and structure evaluations
Task or metricTransfer / language-model representation
Evidence statusPeer-reviewed
Open source ↗

Task-specific evidence only; not comparable as a universal leaderboard score.

Integrated discovery platform

Protein representation and structure evaluations

Version history not yet curated · Split details not yet normalized
peer-reviewed

A structured benchmark claim is recorded; consult the linked source for numeric values and protocol details.

Claim caveats
  • Protocol, split and implementation details must match before comparing this claim with another result.
Open benchmark claim →

Known limitations

  • Performance depends on the evaluation dataset and operating conditions.
  • Task-specific benchmark results should not be compared across unlike domains.
  • Outputs require task-specific scientific and experimental validation.

Milestones

Not normalized

Provides the language-model representation used by ESMFold.