Task

protein-small molecule binding-site prediction

Small-molecule binding-site classification evaluates residue predictions on structural and curated protein annotations.

Sources

Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages

12 evaluations · 45 results

Overview

Datasets

SJC combines sc-PDB, JOINED and COACH420; UniProtSMB supplies a separately curated binding-site dataset.

Metrics

Precision, recall, MCC, AUROC and AUPRC.

Allowed inputs

Protein amino-acid sequences.

Sources

Evaluation procedure diagram

How it worksComputational evaluation flow

Conceptual summary of the cited evaluation; exact task configuration and source version remain part of the protocol.

Sources

limited source coverage · Automated source review, 2026-09-16. All specifications and missing details

Results

Each comparison retains its reviewed evaluation scope, dataset and metric. Results are shown without a pooled ranking.

Evaluation scope and datasetMetric and reported setup

COACH420, trained on CHEN11 · Recall

Recall (unitless) · Higher values are better.

COACH420, trained on CHEN11 (protein-small molecule binding-site prediction) · COACH420, trained on CHEN11

Evidence origin: Result quoted from another source, Author-reported evaluation.

Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Comparison of CLAPE-SMB with other models trained on CHEN11 and tested on COACH420; Table 2 (Tab2), row 2 P2Rank, column 2: COACH420, trained on CHEN11 Recall; Table 2 (Tab2), row 3 GraphBind, column 2: COACH420, trained on CHEN11 Recall; Table 2 (Tab2), row 4 CLAPE-SMB, column 2: COACH420, trained on CHEN11 Recall

Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.

Comparison details and limitations

Residue-level small-molecule binding-site classification. CHEN11 training, COACH420 test.

Automated source review: 2026-09-17. Numerical source review does not establish independent reproduction.

Dots show point estimates. Whiskers show only explicitly defined uncertainty (standard deviation, standard error or a labelled interval); their definitions remain in Table. Unresolved uncertainty is not plotted. Differences do not establish statistical significance.

Find a tested model or method

Showing 3 of 3 matching rows.

Zoom to observed range

Tested configuration

00.250.50.751

Reported score

Methods and evaluation design

Procedure, tasks and evaluated configurations

How it works

Evaluation methodology

SJC combines sc-PDB, JOINED and COACH420; UniProtSMB supplies a separately curated binding-site dataset. UniProtSMB representative proteins are partitioned 80:10:10 for training, validation and testing. Precision, recall, MCC, AUROC and AUPRC. ESM-2 versus ProtBert embeddings and MLP/CNN/Transformer head ablations; some head/model comparisons use the test set. Protein similarity clustering and cluster-aware split comparisons are described; ESM-2 pretraining sequences are not universally excluded. Multiple random seeds are evaluated; the paper reports fold-averaged metrics and standard deviations for its robustness experiment.

Sources

Evaluation design

Benchmarks bring together tasks and protocols. A task describes the biological question; a protocol defines a particular test.

Protocols

These source-backed links do not make different protocols or scores interchangeable.

Recorded evaluations

Each evaluation records what was tested and under which conditions.

CLAPE-SMB with ESM-2: protein-small molecule binding-site prediction

Run this benchmark

Choose a concrete protocol before running an evaluation. Its inputs, split and scoring rules determine which results can be compared.

Run instructions

No runnable recipe has been reviewed for this task. Dataset access, model requirements, licences and compute requirements must be checked against its sources before execution.

A task describes a biological question. Choose a linked protocol to obtain concrete split and scoring instructions.

Strengths, limitations and unresolved questions

Strengths and limitations

Strengths and considerations

No source-reviewed explanatory claims are recorded here yet.

Limitations and conditions

Binding residues are rare and incomplete annotations are discussed in the source. SJC, COACH420 and UniProtSMB results are distinct evaluations, not one shared test set.
Sources
Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages

Profile review details

Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged.

Stable record: reported-task-b181ed450cdd41

Specifications

Inputs, training, access and other details

Explanatory profile: limited source coverage · Automated source review, 2026-09-16. Review applies to the cited claims; unresolved fields are listed below. Numerical results retain their own review status.

Data, procedure and scoring
Property	Description and evidence
Datasets	SJC combines sc-PDB, JOINED and COACH420; UniProtSMB supplies a separately curated binding-site dataset. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages
Splits	UniProtSMB representative proteins are partitioned 80:10:10 for training, validation and testing. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages
Metrics	Precision, recall, MCC, AUROC and AUPRC. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages
Baselines	ESM-2 versus ProtBert embeddings and MLP/CNN/Transformer head ablations; some head/model comparisons use the test set. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages
Leakage controls	Protein similarity clustering and cluster-aware split comparisons are described; ESM-2 pretraining sequences are not universally excluded. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages
Uncertainty	Multiple random seeds are evaluated; the paper reports fold-averaged metrics and standard deviations for its robustness experiment. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages
Entity type	Paper-specific computational evaluation protocol. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages
Organisms	The SJC structural collections and UniProtSMB annotations are selected by binding-site evidence and protein redundancy. Their preparation sections and dataset tables do not summarize organism composition or define a species-specific test. · Not reported in inspected sources Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · SJC dataset preparation; UniProtSMB dataset preparation; Table 1
Assays	Curated protein–small-molecule binding-site annotations. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages
Allowed inputs	Protein amino-acid sequences. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages
Adaptation	Supervised residue-level predictor fitted on the training proteins. Sources Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages

Evidence

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

Papers and result coverage

Last literature check: 2026-09-17. Dated primary-source discovery and protocol/table screening. Source checking does not mean experimental reproduction. Only separately extracted and independently reviewed numeric batches are publishable.

Paper or primary resource	Version	Reference
Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning	version of record	Read source

Historical gaps recorded on 2026-09-17

The catalogue now holds 45 result rows for this benchmark. A note below about pending extraction describes the state on 2026-09-17 and may since have been answered by a later batch. The result rows and their sources are the current record.

exact checkpoint hashes and per-method scored denominators: Table labels alone do not establish these fields; do not infer checkpoint or scored count from model name or dataset size.

Search and extraction details

complete tables extracted

Searches

Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning 10.1186/s13321-024-00920-2

Evidence locations

Table 2; XML table Tab2
Table 3; XML table Tab3
Table 4; XML table Tab4

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.

17 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-09-29-06401fd5b220
Property and statement	Original source and location	Review and provenance
Diagram caption Conceptual summary of the cited evaluation; exact task configuration and source version remain part of the protocol. Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Diagram steps Input: Protein amino-acid sequences. Evaluation: Supervised residue-level predictor fitted on the training proteins. Readout: Precision, recall, MCC, AUROC and AUPRC. Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Diagram title Computational evaluation flow Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Datasets SJC combines sc-PDB, JOINED and COACH420; UniProtSMB supplies a separately curated binding-site dataset. Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.0.value` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Splits UniProtSMB representative proteins are partitioned 80:10:10 for training, validation and testing. Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.1.value` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Adaptation Supervised residue-level predictor fitted on the training proteins. Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.10.value` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Metrics Precision, recall, MCC, AUROC and AUPRC. Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.2.value` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Baselines ESM-2 versus ProtBert embeddings and MLP/CNN/Transformer head ablations; some head/model comparisons use the test set. Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.3.value` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Leakage controls Protein similarity clustering and cluster-aware split comparisons are described; ESM-2 pretraining sequences are not universally excluded. Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.4.value` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Uncertainty Multiple random seeds are evaluated; the paper reports fold-averaged metrics and standard deviations for its robustness experiment. Individual claims	Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning Original source ↗ Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages Version: version of record Retrieved: 2026-09-16T10:41:06Z	source checked automated source review · 2026-09-16 Audit details Relevant full-paper computational evaluation sections, tables/captions and cited supplementary task passages were reviewed. Reporting omissions are scoped to the inspected sources. Original numerical results are unchanged. Field: `attributes.profile.facts.5.value` Source artifact SHA-256: `215919244c3dd2dfb0b55fce91c211430fd8d4aee4bb28bd03eab9f4feb73e62` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact

All evidence (CSV)All evidence (JSONL)How to read the evidence

Sources and history

View linked audit checks and correction history

Release 2026-09-29-06401fd5b220 · Record review: needs review

2 source records and release history

Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Original source · version of record
Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning · Original source · version of record

Download this release

Technical metadata and extraction receipts

Stable ID: reported-task-b181ed450cdd41

areas: proteins-complexes
tasks: protein-small molecule binding-site prediction
entity level: task
version: Not reported
task: protein-small molecule binding-site prediction
scope note: Paper-specific evaluation task; protocol completeness requires further extraction.
comparison panels: id: clape-smb-2024-tab2-recall; title: COACH420, trained on CHEN11 · Recall; protocol id: paper-protocol-efe21a629d215d8f9a; dataset id: paper-dataset-d71026cdf5608821b0; metric: Recall; unit: unitless; direction: higher; result ids: paper-result-cac8b5d6d782329a59; paper-result-519265cd5397a983df; paper-result-8bf20d3347841d493b; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models trained on CHEN11 and tested on COACH420; Table 2 (Tab2), row 2 P2Rank, column 2: COACH420, trained on CHEN11 Recall; Table 2 (Tab2), row 3 GraphBind, column 2: COACH420, trained on CHEN11 Recall; Table 2 (Tab2), row 4 CLAPE-SMB, column 2: COACH420, trained on CHEN11 Recall; context: Residue-level small-molecule binding-site classification. CHEN11 training, COACH420 test.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab2-precision; title: COACH420, trained on CHEN11 · Precision; protocol id: paper-protocol-efe21a629d215d8f9a; dataset id: paper-dataset-d71026cdf5608821b0; metric: Precision; unit: unitless; direction: higher; result ids: paper-result-2b798318eda882028c; paper-result-9000014b7e3080e2bc; paper-result-cef2f0889aade7c724; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models trained on CHEN11 and tested on COACH420; Table 2 (Tab2), row 2 P2Rank, column 3: COACH420, trained on CHEN11 Precision; Table 2 (Tab2), row 3 GraphBind, column 3: COACH420, trained on CHEN11 Precision; Table 2 (Tab2), row 4 CLAPE-SMB, column 3: COACH420, trained on CHEN11 Precision; context: Residue-level small-molecule binding-site classification. CHEN11 training, COACH420 test.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab2-mcc; title: COACH420, trained on CHEN11 · MCC; protocol id: paper-protocol-efe21a629d215d8f9a; dataset id: paper-dataset-d71026cdf5608821b0; metric: MCC; unit: unitless; direction: higher; result ids: paper-result-e198855db4ce0af942; paper-result-7c1331725b140d527d; paper-result-941dc460e719b42566; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models trained on CHEN11 and tested on COACH420; Table 2 (Tab2), row 2 P2Rank, column 4: COACH420, trained on CHEN11 MCC; Table 2 (Tab2), row 3 GraphBind, column 4: COACH420, trained on CHEN11 MCC; Table 2 (Tab2), row 4 CLAPE-SMB, column 4: COACH420, trained on CHEN11 MCC; context: Residue-level small-molecule binding-site classification. CHEN11 training, COACH420 test.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab2-auroc; title: COACH420, trained on CHEN11 · AUROC; protocol id: paper-protocol-efe21a629d215d8f9a; dataset id: paper-dataset-d71026cdf5608821b0; metric: AUROC; unit: unitless; direction: higher; result ids: paper-result-3429737973d4fc729a; paper-result-6dccf13adc1b0316ba; paper-result-061666088fe4993266; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models trained on CHEN11 and tested on COACH420; Table 2 (Tab2), row 2 P2Rank, column 5: COACH420, trained on CHEN11 AUROC; Table 2 (Tab2), row 3 GraphBind, column 5: COACH420, trained on CHEN11 AUROC; Table 2 (Tab2), row 4 CLAPE-SMB, column 5: COACH420, trained on CHEN11 AUROC; context: Residue-level small-molecule binding-site classification. CHEN11 training, COACH420 test.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab3-recall; title: SJC test split · Recall; protocol id: paper-protocol-07cf559d2f1f9262e7; dataset id: paper-dataset-623c309e2fb4d39ef8; metric: Recall; unit: unitless; direction: higher; result ids: paper-result-3c8603f3a617ae353d; paper-result-b455cc6c453819bf59; paper-result-86b3e09746a262c729; paper-result-f678244aa5db050c82; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models on the SJC; Table 3 (Tab3), row 2 P2Rank, column 2: SJC test split Recall; Table 3 (Tab3), row 3 GraphBind, column 2: SJC test split Recall; Table 3 (Tab3), row 4 DeepProSite, column 2: SJC test split Recall; Table 3 (Tab3), row 5 CLAPE-SMB, column 2: SJC test split Recall; context: Residue-level small-molecule binding-site classification. 80/10/10 train/validation/test partition of nonredundant sequences.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab3-precision; title: SJC test split · Precision; protocol id: paper-protocol-07cf559d2f1f9262e7; dataset id: paper-dataset-623c309e2fb4d39ef8; metric: Precision; unit: unitless; direction: higher; result ids: paper-result-c76919342c3926a7e8; paper-result-c04870d60a021ac0ed; paper-result-fcf6afead98be8b7bc; paper-result-02b4a5a19b61ad5217; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models on the SJC; Table 3 (Tab3), row 2 P2Rank, column 3: SJC test split Precision; Table 3 (Tab3), row 3 GraphBind, column 3: SJC test split Precision; Table 3 (Tab3), row 4 DeepProSite, column 3: SJC test split Precision; Table 3 (Tab3), row 5 CLAPE-SMB, column 3: SJC test split Precision; context: Residue-level small-molecule binding-site classification. 80/10/10 train/validation/test partition of nonredundant sequences.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab3-mcc; title: SJC test split · MCC; protocol id: paper-protocol-07cf559d2f1f9262e7; dataset id: paper-dataset-623c309e2fb4d39ef8; metric: MCC; unit: unitless; direction: higher; result ids: paper-result-63c3e75a69cfad040e; paper-result-42d845442e40d9846c; paper-result-bafeb53a23564fc9e5; paper-result-936e56c8fad08e6191; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models on the SJC; Table 3 (Tab3), row 2 P2Rank, column 4: SJC test split MCC; Table 3 (Tab3), row 3 GraphBind, column 4: SJC test split MCC; Table 3 (Tab3), row 4 DeepProSite, column 4: SJC test split MCC; Table 3 (Tab3), row 5 CLAPE-SMB, column 4: SJC test split MCC; context: Residue-level small-molecule binding-site classification. 80/10/10 train/validation/test partition of nonredundant sequences.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab3-auroc; title: SJC test split · AUROC; protocol id: paper-protocol-07cf559d2f1f9262e7; dataset id: paper-dataset-623c309e2fb4d39ef8; metric: AUROC; unit: unitless; direction: higher; result ids: paper-result-a1b671f27b279cb8b3; paper-result-afdce2cee34b41cf09; paper-result-87d8d792a85dd7223d; paper-result-a337f472f62cd0d12a; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models on the SJC; Table 3 (Tab3), row 2 P2Rank, column 5: SJC test split AUROC; Table 3 (Tab3), row 3 GraphBind, column 5: SJC test split AUROC; Table 3 (Tab3), row 4 DeepProSite, column 5: SJC test split AUROC; Table 3 (Tab3), row 5 CLAPE-SMB, column 5: SJC test split AUROC; context: Residue-level small-molecule binding-site classification. 80/10/10 train/validation/test partition of nonredundant sequences.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab4-recall; title: UniProtSMB test split · Recall; protocol id: paper-protocol-89b76bfabd1dfe31a6; dataset id: paper-dataset-7ed13bec749f1e245b; metric: Recall; unit: unitless; direction: higher; result ids: paper-result-a52e839c67a23202a9; paper-result-1fef4fca1431b14c73; paper-result-08de771b9d631cbb90; paper-result-114f60a38e97139e57; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models on the UniProtSMB; Table 4 (Tab4), row 2 P2Rank, column 2: UniProtSMB test split Recall; Table 4 (Tab4), row 3 GraphBind, column 2: UniProtSMB test split Recall; Table 4 (Tab4), row 4 DeepProSite, column 2: UniProtSMB test split Recall; Table 4 (Tab4), row 5 CLAPE-SMB, column 2: UniProtSMB test split Recall; context: Residue-level small-molecule binding-site classification. 80/10/10 train/validation/test partition of nonredundant sequences.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab4-precision; title: UniProtSMB test split · Precision; protocol id: paper-protocol-89b76bfabd1dfe31a6; dataset id: paper-dataset-7ed13bec749f1e245b; metric: Precision; unit: unitless; direction: higher; result ids: paper-result-9d84f8da0854369c5c; paper-result-ba1e3a15b41c7f68e7; paper-result-a1f27fb075923f53e8; paper-result-0cb8ce6c55a8f0597f; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models on the UniProtSMB; Table 4 (Tab4), row 2 P2Rank, column 3: UniProtSMB test split Precision; Table 4 (Tab4), row 3 GraphBind, column 3: UniProtSMB test split Precision; Table 4 (Tab4), row 4 DeepProSite, column 3: UniProtSMB test split Precision; Table 4 (Tab4), row 5 CLAPE-SMB, column 3: UniProtSMB test split Precision; context: Residue-level small-molecule binding-site classification. 80/10/10 train/validation/test partition of nonredundant sequences.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab4-mcc; title: UniProtSMB test split · MCC; protocol id: paper-protocol-89b76bfabd1dfe31a6; dataset id: paper-dataset-7ed13bec749f1e245b; metric: MCC; unit: unitless; direction: higher; result ids: paper-result-7ed2311155bd397324; paper-result-de34ece35da8021e97; paper-result-ba078e213fdac2e0bf; paper-result-90ff2ad95b57c8f0f6; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models on the UniProtSMB; Table 4 (Tab4), row 2 P2Rank, column 4: UniProtSMB test split MCC; Table 4 (Tab4), row 3 GraphBind, column 4: UniProtSMB test split MCC; Table 4 (Tab4), row 4 DeepProSite, column 4: UniProtSMB test split MCC; Table 4 (Tab4), row 5 CLAPE-SMB, column 4: UniProtSMB test split MCC; context: Residue-level small-molecule binding-site classification. 80/10/10 train/validation/test partition of nonredundant sequences.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17; id: clape-smb-2024-tab4-auroc; title: UniProtSMB test split · AUROC; protocol id: paper-protocol-89b76bfabd1dfe31a6; dataset id: paper-dataset-7ed13bec749f1e245b; metric: AUROC; unit: unitless; direction: higher; result ids: paper-result-7011ff11e364052ff1; paper-result-195480a469229de195; paper-result-d0fa42d3c13fe8ec4d; paper-result-e4602df92496bbb342; source ids: clape-smb-2024; source locator: Comparison of CLAPE-SMB with other models on the UniProtSMB; Table 4 (Tab4), row 2 P2Rank, column 5: UniProtSMB test split AUROC; Table 4 (Tab4), row 3 GraphBind, column 5: UniProtSMB test split AUROC; Table 4 (Tab4), row 4 DeepProSite, column 5: UniProtSMB test split AUROC; Table 4 (Tab4), row 5 CLAPE-SMB, column 5: UniProtSMB test split AUROC; context: Residue-level small-molecule binding-site classification. 80/10/10 train/validation/test partition of nonredundant sequences.; caveats: Sequence-only and structure-using predictors have unequal input information and are labelled method comparisons, not architecture-controlled tests. CLAPE-SMB seeds 6,17,35,42; the table prints ± but does not explicitly label the spread in its caption. N/A AUROC remains unavailable. Origins are labelled per method; appearance in one table does not constitute independent replication of every model.; review: method: automated_source_review; date: 2026-09-17
benchmark research: review date: 2026-09-17; status: complete_tables_extracted; primary sources: evidence-expansion-clape-smb-2024-21591924; inspected locators: Table 2; XML table Tab2; Table 3; XML table Tab3; Table 4; XML table Tab4; searched queries: Protein-small molecule binding site prediction based on a pre-trained protein language model with contrastive learning 10.1186/s13321-024-00920-2; gaps: exact checkpoint hashes and per-method scored denominators: Table labels alone do not establish these fields; do not infer checkpoint or scored count from model name or dataset size.; claim scope: Dated primary-source discovery and protocol/table screening. Source checking does not mean experimental reproduction. Only separately extracted and independently reviewed numeric batches are publishable.
historical missing metadata: protocol version: not_reported_in_legacy_extract; split: not_reported_in_legacy_extract
metadata review scope: historical_missing_metadata preserves the original discovery state. Current descriptive evidence and missingness are recorded in profile.facts; numerical-result review is separate.
legacy kinds: benchmark
entity classification: review date: 2026-09-17; rationale: This source-scoped record identifies the biological prediction task and holds its paper context. Preserve the existing task identity; exact split, model adaptation and scoring remain in linked evaluations or separate protocol records.; source ids: clape-smb-2024; source locator: Methods: Evaluation metrics; SJC dataset preparation; UniProtSMB dataset preparation; Discussion; cached text lines 24–25, 35–43, 93; comparative evaluation and ablation passages; uncertainty/repeat-run/statistical-comparison passages; ambiguities: A paper- or suite-specific task may constrain some inputs or metrics; that alone does not make it interchangeable with a complete versioned protocol. No protocol equivalence is inferred.; Some legacy profile Entity type facts use the generic phrase computational evaluation protocol. That boilerplate is not sufficient to establish a single fixed protocol identity or to merge this task with another protocol record.

Related records

dataset: SJC
benchmark: CLAPE-SMB with ESM-2: protein-small molecule binding-site prediction
evaluates task: SJC test split (protein-small molecule binding-site prediction)
evaluates task: UniProtSMB test split (protein-small molecule binding-site prediction)
evaluates task: COACH420, trained on CHEN11 (protein-small molecule binding-site prediction)

Suggest a correction