Configuration

Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)

Configuration of Boltz-2: Published checkpoint applied without retraining; co-folds its own complex from sequence and SMILES, so the supplied structures are not consumed.

3 evaluations · 19 results

Overview

An explanatory profile has not yet been reviewed for this record.

Consult the linked sources for architecture or protocol details. Missing evidence is not evidence of a missing capability.

Evaluations and results

3 evaluations · 19 results. Different protocols are not a single leaderboard.

Sorted by Concordance index (PLABench casp16-L1000) (higher is better). The best value in each column is highlighted. Decimals are rounded for display; each value links to the printed value and its source.

Tested configuration	Protocol and dataset	Concordance index (PLABench casp16-L1000)	Concordance index (PLABench casp16-L3000)	Concordance index (PLABench chembl35)	Kendall tau (PLABench chembl35)	Modified squared correlation (PLABench casp16-L1000)	Modified squared correlation (PLABench casp16-L3000)	Modified squared correlation (PLABench chembl35)	MSE (PLABench casp16-L1000)	MSE (PLABench casp16-L3000)	MSE (PLABench chembl35)	Pearson (PLABench casp16-L1000)	Pearson (PLABench casp16-L3000)	Pearson (PLABench chembl35)	RMSE (PLABench casp16-L1000)	RMSE (PLABench casp16-L3000)	RMSE (PLABench chembl35)	Spearman (PLABench casp16-L1000)	Spearman (PLABench casp16-L3000)	Spearman (PLABench chembl35)
Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)Evaluation	PLABench CASP16 L1000 affinity, CASP16 ligand affinity series L1000 (chymase)	0.618	Not reported	Not reported	Not reported	0.068	Not reported	Not reported	0.436	Not reported	Not reported	0.544	Not reported	Not reported	0.66	Not reported	Not reported	0.382	Not reported	Not reported
Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)Evaluation	PLABench CASP16 L3000 affinity, CASP16 ligand affinity series L3000 (autotaxin)	Not reported	0.745	Not reported	Not reported	Not reported	0.356	Not reported	Not reported	1.03	Not reported	Not reported	0.67	Not reported	Not reported	1.02	Not reported	Not reported	0.67	Not reported
Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)Evaluation	PLABench ChEMBL35 leakage-controlled affinity, PLABench ChEMBL35 leakage-controlled test set	Not reported	Not reported	0.606	0.211	Not reported	Not reported	0.098	Not reported	Not reported	1.69	Not reported	Not reported	0.321	Not reported	Not reported	1.25	Not reported	Not reported	0.302

All 19 result rows with coverage, uncertainty and sources

Filter evaluations

Tested model or methodProtocolDatasetEvaluation setupMetricEvidence origin

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configuration	Protocol and dataset	Finding	Evidence and details
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L1000 affinity Dataset: CASP16 ligand affinity series L1000 (chymase)	0.618 Concordance index (PLABench casp16-L1000) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L1000 lit-interactions-20261011-leakctl-protocol-casp16-l1000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L1000', column 'CI' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L1000 affinity Dataset: CASP16 ligand affinity series L1000 (chymase)	0.436 MSE (PLABench casp16-L1000) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L1000 lit-interactions-20261011-leakctl-protocol-casp16-l1000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L1000', column 'MSE' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L1000 affinity Dataset: CASP16 ligand affinity series L1000 (chymase)	0.544 Pearson (PLABench casp16-L1000) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L1000 lit-interactions-20261011-leakctl-protocol-casp16-l1000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L1000', column 'Pearson' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L1000 affinity Dataset: CASP16 ligand affinity series L1000 (chymase)	0.068 Modified squared correlation (PLABench casp16-L1000) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L1000 lit-interactions-20261011-leakctl-protocol-casp16-l1000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L1000', column 'rm2' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L1000 affinity Dataset: CASP16 ligand affinity series L1000 (chymase)	0.66 RMSE (PLABench casp16-L1000) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L1000 lit-interactions-20261011-leakctl-protocol-casp16-l1000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L1000', column 'RMSE' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L1000 affinity Dataset: CASP16 ligand affinity series L1000 (chymase)	0.382 Spearman (PLABench casp16-L1000) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L1000 lit-interactions-20261011-leakctl-protocol-casp16-l1000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L1000', column 'Spearman' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L3000 affinity Dataset: CASP16 ligand affinity series L3000 (autotaxin)	0.745 Concordance index (PLABench casp16-L3000) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L3000 lit-interactions-20261011-leakctl-protocol-casp16-l3000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L3000', column 'CI' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L3000 affinity Dataset: CASP16 ligand affinity series L3000 (autotaxin)	1.03 MSE (PLABench casp16-L3000) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L3000 lit-interactions-20261011-leakctl-protocol-casp16-l3000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L3000', column 'MSE' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L3000 affinity Dataset: CASP16 ligand affinity series L3000 (autotaxin)	0.67 Pearson (PLABench casp16-L3000) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L3000 lit-interactions-20261011-leakctl-protocol-casp16-l3000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L3000', column 'Pearson' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L3000 affinity Dataset: CASP16 ligand affinity series L3000 (autotaxin)	0.356 Modified squared correlation (PLABench casp16-L3000) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L3000 lit-interactions-20261011-leakctl-protocol-casp16-l3000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L3000', column 'rm2' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L3000 affinity Dataset: CASP16 ligand affinity series L3000 (autotaxin)	1.02 RMSE (PLABench casp16-L3000) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L3000 lit-interactions-20261011-leakctl-protocol-casp16-l3000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L3000', column 'RMSE' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench CASP16 L3000 affinity Dataset: CASP16 ligand affinity series L3000 (autotaxin)	0.67 Spearman (PLABench casp16-L3000) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench casp16-L3000 lit-interactions-20261011-leakctl-protocol-casp16-l3000 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 3, row 'Boltz2', series 'L3000', column 'Spearman' (text-layer line 832)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench ChEMBL35 leakage-controlled affinity Dataset: PLABench ChEMBL35 leakage-controlled test set	0.606 Concordance index (PLABench chembl35) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench chembl35 lit-interactions-20261011-leakctl-protocol-chembl35 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 4, row 'Boltz2', column 'CI' (text-layer line 1032)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench ChEMBL35 leakage-controlled affinity Dataset: PLABench ChEMBL35 leakage-controlled test set	0.211 Kendall tau (PLABench chembl35) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench chembl35 lit-interactions-20261011-leakctl-protocol-chembl35 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 4, row 'Boltz2', column 'Kendall' (text-layer line 1032)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench ChEMBL35 leakage-controlled affinity Dataset: PLABench ChEMBL35 leakage-controlled test set	1.69 MSE (PLABench chembl35) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench chembl35 lit-interactions-20261011-leakctl-protocol-chembl35 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 4, row 'Boltz2', column 'MSE' (text-layer line 1032)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench ChEMBL35 leakage-controlled affinity Dataset: PLABench ChEMBL35 leakage-controlled test set	0.321 Pearson (PLABench chembl35) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench chembl35 lit-interactions-20261011-leakctl-protocol-chembl35 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 4, row 'Boltz2', column 'Pearson' (text-layer line 1032)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench ChEMBL35 leakage-controlled affinity Dataset: PLABench ChEMBL35 leakage-controlled test set	0.098 Modified squared correlation (PLABench chembl35) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench chembl35 lit-interactions-20261011-leakctl-protocol-chembl35 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 4, row 'Boltz2', column 'rm2' (text-layer line 1032)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench ChEMBL35 leakage-controlled affinity Dataset: PLABench ChEMBL35 leakage-controlled test set	1.25 RMSE (PLABench chembl35) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench chembl35 lit-interactions-20261011-leakctl-protocol-chembl35 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 4, row 'Boltz2', column 'RMSE' (text-layer line 1032)
Configuration: Boltz2 (CASP16 and ChEMBL35 evaluation, PLABench)	Protocol: PLABench ChEMBL35 leakage-controlled affinity Dataset: PLABench ChEMBL35 leakage-controlled test set	0.302 Spearman (PLABench chembl35) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Boltz2 on PLABench chembl35 lit-interactions-20261011-leakctl-protocol-chembl35 Aggregation: Not reported Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Table 4, row 'Boltz2', column 'Spearman' (text-layer line 1032)

Source checking is not independent reproduction. Release 2026-10-23-5347695500ed.

Use this configuration

How it works, versions and access

Additional methodological details, versions and access terms have not yet been extracted into a structured profile for this configuration. Consult the linked sources and evaluation records below.

Evidence

Source checking verifies the cited claim or transcription. It does not establish independent reproduction.

Evidence table

Inspect claims, sources and review details

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.

0 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-10-23-5347695500ed
Property and statement	Original source and location	Review and provenance

No evidence rows match these filters. Choose another scope or clear the search.

How to read the evidence

Sources and history

Release 2026-10-23-5347695500ed · Record review: source checked

1 source record and release history

Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction · Original source · bioRxiv 10.64898/2026.09.17.752433v1 (posted 2026-09-23); 30-page PDF

Download this release (gzip)

Technical metadata and extraction receipts

Stable ID: lit-interactions-20261011-leakctl-config-boltz2-casp16-chembl35

areas: molecular-interactions
contexts: research
method types: foundation_model
reported name: Boltz2
foundation model eligible: true
parameters: Published checkpoint applied without retraining; co-folds its own complex from sequence and SMILES, so the supplied structures are not consumed. 10 structural samples, affinity scored on the highest-ipTM pose, 500 sampling steps, 5 recycling steps (Methods lines 437-439)
source locator: Table 1; Methods; Tables 3 and 4
access reuse: Not stated per model in the source; see each model's release
missing metadata: version: reason: unreported; note: No checkpoint version printed
provenance note: Boltz2 as evaluated in PLABench (structure-based).

Related records

Suggest a correction