ESM2 8M: Held-out human-versus-virus protein classification
Compare protein origin classifiers; T 5 embeddings plus linear/tree models vs ESM 2 fine-tuning and simple controls. UniRef90 deduplication; proteins longer than1,600 residues excluded; UniRef50 clusters assigned80% training and20% test with no cluster shared. Separate four-fold error-analysis experiment not assigned toTable1.
Evaluation procedure
Compare protein origin classifiers; T 5 embeddings plus linear/tree models vs ESM 2 fine-tuning and simple controls. UniRef90 deduplication; proteins longer than1,600 residues excluded; UniRef50 clusters assigned80% training and20% test with no cluster shared. Separate four-fold error-analysis experiment not assigned toTable1.
- Configuration
- ESM2 8M
- Protocol
- Held-out human-versus-virus protein classification (human-versus-viral protein classification)
- Dataset
- human and viral proteins
- origin
- Author-reported evaluation
- configuration
- Paper-specific evaluated pipeline; exact checkpoint not inferred from label
- protocol id
- paper-protocol-1352834b9391c1dacb
- dataset version
- Not reported
- split
- UniRef90 deduplication; proteins longer than1,600 residues excluded; UniRef50 clusters assigned80% training and20% test with no cluster shared. Separate four-fold error-analysis experiment not assigned toTable1.
- subset
- Held-out human-versus-virus protein classification
- population
- Not reported
- aggregation
- Precision and recall macro-averaged over classes; AUROC and log-loss computed with scikit-learn.
- inputs
- Not reported
- adaptation
- Paper-specific evaluated pipeline; exact checkpoint not inferred from label
- budget
- Not reported
- metric implementation
- Not reported
Metadata review: needs review. Unreported conditions prevent automatic comparisons.
Evaluation results
Release 2026-09-17-d277315f7d76 · 1 evaluation · 4 metric rows. Different protocols are not a single leaderboard.
| Metric and finding | Coverage and uncertainty | Evidence |
|---|---|---|
| ESM2 8M: Held-out human-versus-virus protein classification Configuration: ESM2 8MProtocol: Held-out human-versus-virus protein classification (human-versus-viral protein classification)Dataset: human and viral proteins Compare protein origin classifiers; T 5 embeddings plus linear/tree models vs ESM 2 fine-tuning and simple controls. UniRef90 deduplication; proteins longer than1,600 residues excluded; UniRef50 clusters assigned80% training and20% test with no cluster shared. Separate four-fold error-analysis experiment not assigned toTable1. Author-reported evaluation · Evaluation metadata: needs review | ||
| 92.15% Prec. Unit: percent · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedProtein Language Models Expose Viral Immune Mimicry · Table 1, row ESM2 8M, column Prec.; XML row4 column4 Source checking is not independent reproduction. |
| 92.33% Recall Unit: percent · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedProtein Language Models Expose Viral Immune Mimicry · Table 1, row ESM2 8M, column Recall; XML row4 column5 Source checking is not independent reproduction. |
| 98.09% AUROC Unit: percent · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedProtein Language Models Expose Viral Immune Mimicry · Table 1, row ESM2 8M, column AUC (%); XML row4 column2 Source checking is not independent reproduction. |
| 94.72% Accur. Unit: percent · Direction: higher | Uncertainty: unreported Scored: Not reported · Eligible: Not reported | source checkedProtein Language Models Expose Viral Immune Mimicry · Table 1, row ESM2 8M, column Accur.; XML row4 column3 Source checking is not independent reproduction. |
Evidence table
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
19 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| attributes.comparison.adaptation Paper-specific evaluated pipeline; exact checkpoint not inferred from label Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | not individually reviewed No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.comparison.aggregation Precision and recall macro-averaged over classes; AUROC and log-loss computed with scikit-learn. Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | not individually reviewed No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.comparison.budget No value recorded Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | missing or unspecified No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.comparison.dataset_version No value recorded Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | missing or unspecified No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.comparison.inputs No value recorded Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | missing or unspecified No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.comparison.metric_implementation No value recorded Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | missing or unspecified No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.comparison.population No value recorded Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | missing or unspecified No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.comparison.protocol_id paper-protocol-1352834b9391c1dacb Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | not individually reviewed No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.comparison.split UniRef90 deduplication; proteins longer than1,600 residues excluded; UniRef50 clusters assigned80% training and20% test with no cluster shared. Separate four-fold error-analysis experiment not assigned toTable1. Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | not individually reviewed No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
| attributes.comparison.subset Held-out human-versus-virus protein classification Context-only references | Protein Language Models Expose Viral Immune Mimicry Table 1: AUC (%), Held-out human-versus-virus protein classification Version: version of record | not individually reviewed No individual claim review recorded author reported Audit detailsField: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. |
Sources and history
Release 2026-09-17-d277315f7d76 · Record review: needs review
- Protein Language Models Expose Viral Immune Mimicry · Original source · version of record
Technical metadata and extraction receipts
Stable ID: paper-evaluation-8a89666251b30349ea
- areas
- proteins-complexes
- tasks
- human-versus-viral protein classification
- origin
- author_reported
- protocol
- Compare protein origin classifiers; T 5 embeddings plus linear/tree models vs ESM 2 fine-tuning and simple controls. UniRef90 deduplication; proteins longer than1,600 residues excluded; UniRef50 clusters assigned80% training and20% test with no cluster shared. Separate four-fold error-analysis experiment not assigned toTable1.
- version
- Paper-specific evaluated pipeline; exact checkpoint not inferred from label
- comparison
- protocol id: paper-protocol-1352834b9391c1dacb; dataset version: Not reported; split: UniRef90 deduplication; proteins longer than1,600 residues excluded; UniRef50 clusters assigned80% training and20% test with no cluster shared. Separate four-fold error-analysis experiment not assigned toTable1.; subset: Held-out human-versus-virus protein classification; population: Not reported; aggregation: Precision and recall macro-averaged over classes; AUROC and log-loss computed with scikit-learn.; inputs: Not reported; adaptation: Paper-specific evaluated pipeline; exact checkpoint not inferred from label; budget: Not reported; metric implementation: Not reported
- source locator
- Table 1: AUC (%), Held-out human-versus-virus protein classification
- missing metadata
- checkpoint revision: unextracted; budget: unreported; split manifest: unextracted
Related records
- model: ESM2 8M
- benchmark: Held-out human-versus-virus protein classification (human-versus-viral protein classification)
- dataset: human and viral proteins
- evaluation: ESM2 8M: Prec. on Held-out human-versus-virus protein classification
- evaluation: ESM2 8M: Recall on Held-out human-versus-virus protein classification
- evaluation: ESM2 8M: AUROC on Held-out human-versus-virus protein classification
- evaluation: ESM2 8M: Accur. on Held-out human-versus-virus protein classification