suite · audited-with-caveats · verified 2026-08-20

AbBiBench

A framework using antibody–antigen complexes to evaluate affinity prediction and antibody redesign.

Audited with caveats: 2 field(s) are marked provisional or conflicted. Warnings are shown next to affected values and these claims are excluded from unqualified comparisons.

Benchmark definition

What is counted

Version: initial-release
Total: 184500 (Standardized benchmark data compiling 184,500 mutated antibodies.)
Task formats: unclassified
Capabilities: PredictionDesignGenerationOptimization
Modalities: Protein sequence3D structureWet-lab output

Version history

Version	Status	Release / as-of	Total	Formal tracks
initial-release `abbibench-initial-release-version`	current	2025-05-23	184500 (Standardized benchmark data compiling 184,500 mutated antibodies.)	None registered

Scientific Task Atlas

Scientific task classification

partial for initial-release. No Scientific Task claim passed independent high-confidence verification; task mapping remains pending a targeted official-source audit.

The source names only a broad direction; no more specific leaf task can be assigned without inference.

Relationship registry

How works use this benchmark

Partial claims, non-evaluation uses, and third-party summaries stay visible without entering model comparisons.

Partial evaluation claims

evaluation

abbibench-a-benchmark-for-antibody-binding-affinity-ma-abbibench-2-use

Work: AbBiBench: A Benchmark for Antibody Binding Affinity Maturation and Design · source version abbibench-a-benchmark-for-antibody-binding-affinity-ma-arxiv-v2

Selection: not reported
Models: ESM-IF1, AntiBERTy, AntiFold, CurrAb, DiffAb, DiffAb fixbb, dyMEAN, dyMEAN fixbb, ESM2, ESM3, MEAN, MEAN fixbb, ProGen2-large, ProSST, SaProt, ProteinMPNN, ProtGPT2
Metrics: Not reported / not applicable
Linked runs: None

Not reported / unresolved: Exact model versions, providers, seeds, repeats, and confidence intervals are not reported.; The source conflicts on light-chain inclusion and on the 1mlc evaluation count.; Exact model and tool versions, compute time, and confidence intervals are not reported.; DiffAb seeds are described only as reaching up to 15; the exact seed list is absent.; The numeric ELISA detection threshold is not reported.; The final Pareto-candidate count conflicts between 18 and 21.; Exact model versions, seeds, repeats, and exact p-values are not reported.; benchmark version; realized n/scope; metric; numeric result; prompt and tools; grader and repeats

Owner-reviewed conservative publication: the creator evaluation is retained only as a partial relationship; conflicted settings and outcomes are omitted pending manual reconciliation.

Evidence

section: Section 3
Supports: /relation_type
section: Section 3
Supports: /benchmark_id
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
figure: Figure 3
Supports: /model_ids
table: Table S2
Supports: /model_ids
other: Table S2 and registry-context.json model proteinmpnn
Supports: /model_ids
figure: Figure 3
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids
table: Table S2
Supports: /model_ids

evaluation

explainable-protein-protein-binding-affinity-predictio-abbibench-9-use

Work: Explainable protein-protein binding affinity prediction via fine-tuning protein language models · source version explainable-protein-protein-binding-affinity-predictio-biorxiv-version-posted-2

Selection: filtered · Nine selected DMS assays: 1n8z, 1mhp_LC, 3gbn_h1, 3gbn_h9, 4fqi_h1, aayl49_ml, aayl50_LC, aayl51, aayl52
Models: BALM-PPI
Metrics: RMSE
Linked runs: None

Not reported / unresolved: Benchmark version is not reported.; The source reports assay inventories, but not an exact realized evaluation n for every assay.; BALM-PPI provider and version are not explicitly reported.; benchmark version

AI-assisted double-pass extraction; n=9 counts selected assay tracks, not individual examples. Values are limited to independently supported claims.

Evidence

section: 2.4
Supports: /relation_type
section: 2.4
Supports: /benchmark_id
section: 2.4
Supports: /scope
section: 2.4
Supports: /scope
table: Table S5
Supports: /scope
table: Table S5
Supports: /scope
table: Table S6
Supports: /model_ids
table: Table S6
Supports: /metric_labels

Creation, training, validation, or model-selection uses

benchmark creation

abbibench-a-benchmark-for-antibody-binding-affinity-ma-abbibench-1-use

Work: AbBiBench: A Benchmark for Antibody Binding Affinity Maturation and Design · source version abbibench-a-benchmark-for-antibody-binding-affinity-ma-arxiv-v2

Selection: not applicable
Models: Not reported / not applicable
Metrics: Not reported / not applicable
Linked runs: None

Not reported / unresolved: A benchmark version and artifact release date are not reported.; The source conflicts on whether AbBiBench contains 14 or 16 binding-affinity assays.

AI-assisted double-pass extraction; values are limited to independently supported claims.

Evidence

section: Introduction
Supports: /relation_type
section: Introduction
Supports: /benchmark_id

fine tuning

explainable-protein-protein-binding-affinity-predictio-abbibench-10-use

Selection: not applicable · Few-Shot 10%
Models: BALM-PPI
Metrics: RMSE
Linked runs: None

Not reported / unresolved: Benchmark version and exact realized training/evaluation n per assay are not reported.; BALM-PPI provider and version are not explicitly reported.

AI-assisted double-pass extraction; values are limited to independently supported claims.

Evidence

table: Table S6
Supports: /relation_type
table: Table S6
Supports: /benchmark_id
section: 2.4
Supports: /scope
section: 2.4
Supports: /scope
table: Table S5
Supports: /scope
table: Table S6
Supports: /scope
table: Table S6
Supports: /model_ids
table: Table S6
Supports: /metric_labels

fine tuning

explainable-protein-protein-binding-affinity-predictio-abbibench-11-use

Selection: not applicable · Few-Shot 20%
Models: BALM-PPI
Metrics: RMSE
Linked runs: None

Not reported / unresolved: Benchmark version and exact realized training/evaluation n per assay are not reported.; BALM-PPI provider and version are not explicitly reported.

AI-assisted double-pass extraction; values are limited to independently supported claims.

Evidence

table: Table S6
Supports: /relation_type
table: Table S6
Supports: /benchmark_id
section: 2.4
Supports: /scope
section: 2.4
Supports: /scope
table: Table S5
Supports: /scope
table: Table S6
Supports: /scope
table: Table S6
Supports: /model_ids
table: Table S6
Supports: /metric_labels

fine tuning

explainable-protein-protein-binding-affinity-predictio-abbibench-12-use

Selection: not applicable · Few-Shot 30%
Models: BALM-PPI
Metrics: RMSE
Linked runs: None

Not reported / unresolved: Benchmark version and exact realized training/evaluation n per assay are not reported.; BALM-PPI provider and version are not explicitly reported.

AI-assisted double-pass extraction; values are limited to independently supported claims.

Evidence

table: Table S6
Supports: /relation_type
table: Table S6
Supports: /benchmark_id
section: 2.4
Supports: /scope
section: 2.4
Supports: /scope
table: Table S5
Supports: /scope
table: Table S6
Supports: /scope
table: Table S6
Supports: /model_ids
table: Table S6
Supports: /metric_labels

Evaluation registry

Works and run settings

A setting change—scope, prompt, tools, budget, grader, or repeats—creates a separate run. Charts never cross a comparability group.

No normalized evaluation run is published yet. Creator evidence is still attached below.

Evidence and change history

Source locators remain visible; expand an item to inspect the exact Registry fields it supports.

AbBiBench: A Benchmark for Antibody Binding Affinity Maturation and Design · section: Supplement 1 · Supports 2 fields

What is counted

Version history

Scientific task classification

How works use this benchmark

Partial evaluation claims

abbibench-a-benchmark-for-antibody-binding-affinity-ma-abbibench-2-use

explainable-protein-protein-binding-affinity-predictio-abbibench-9-use

Creation, training, validation, or model-selection uses

abbibench-a-benchmark-for-antibody-binding-affinity-ma-abbibench-1-use

explainable-protein-protein-binding-affinity-predictio-abbibench-10-use

explainable-protein-protein-binding-affinity-predictio-abbibench-11-use

explainable-protein-protein-binding-affinity-predictio-abbibench-12-use

Works and run settings

Evidence and change history

Unresolved field claims