Protocol

Davis drug-target affinity, DeepDTA holdout fold

Affinity regression on the Davis DeepDTA holdout fold; CI and MSE as printed.

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.

16 recorded evaluations, 32 metric rows. A comparison chart has not yet been validated for these results. The table retains the individual findings and their sources.

View coverage and remaining gaps across all benchmarks

Results

Results are available, but no reviewed comparison panel is linked in this release.

All evaluations

16 evaluations · 32 results. Different protocols are not a single leaderboard.

Protocol: Davis drug-target affinity, DeepDTA holdout fold · Dataset: Davis holdout test fold (DeepDTA split)

Sorted by Concordance index (Davis DeepDTA holdout fold) (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	Concordance index (Davis DeepDTA holdout fold)	MSE (Davis DeepDTA holdout fold)
PLM-GraphDTA with ESMC 600M protein and GAT-GCN drug representationEvaluation	0.91 (best in column)	0.169
PLM-GraphDTA with ESMC 600M protein and GIN drug representationEvaluation	0.91 (best in column)	0.168
DeepGraphDTA with 1D conv. (32) protein and GIN drug representationEvaluation	0.909	0.172
PLM-GraphDTA with ESMC 600M protein and GAT drug representationEvaluation	0.908	0.174
DeepGraphDTA with 1D conv. (16) protein and GAT drug representationEvaluation	0.907	0.176
DeepGraphDTA with 1D conv. (16) protein and GCN drug representationEvaluation	0.906	0.174
DeepGraphDTA with 1D conv. (8) protein and GAT-GCN drug representationEvaluation	0.906	0.175
PLM-GraphDTA with ESMC 600M protein and GCN drug representationEvaluation	0.906	0.167 (best in column)
DeepDTA with 1D conv. protein and 1D conv. drug representationEvaluation	0.902	0.224
PGraphDTA with ESM2 650M protein and GAT drug representationEvaluation	0.898	0.206
GraphDTA with 1D conv. protein and GIN drug representationEvaluation	0.897	0.199
GraphDTA with 1D conv. protein and GAT drug representationEvaluation	0.893	0.255
GraphDTA with 1D conv. protein and GAT-GCN drug representationEvaluation	0.889	0.227
SimBoost with Similarity features protein and Similarity features drug representationEvaluation	0.885	0.246
GraphDTA with 1D conv. protein and GCN drug representationEvaluation	0.88	0.237
KronRLS with Similarity vector protein and Similarity vector drug representationEvaluation	0.87	0.359

All 32 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: DeepDTA with 1D conv. protein and 1D conv. drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.902 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepDTA (1D conv.; 1D conv.) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepDTA' with protein '1D conv.' and drug '1D conv.', column 'CI' (text-layer line 529)
Configuration: DeepDTA with 1D conv. protein and 1D conv. drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.224 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepDTA (1D conv.; 1D conv.) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepDTA' with protein '1D conv.' and drug '1D conv.', column 'MSE' (text-layer line 529)
Configuration: DeepGraphDTA with 1D conv. (16) protein and GAT drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.907 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepGraphDTA (1D conv. (16); GAT) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepGraphDTA' with protein '1D conv. (16)' and drug 'GAT', column 'CI' (text-layer line 534)
Configuration: DeepGraphDTA with 1D conv. (16) protein and GAT drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.176 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepGraphDTA (1D conv. (16); GAT) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepGraphDTA' with protein '1D conv. (16)' and drug 'GAT', column 'MSE' (text-layer line 534)
Configuration: DeepGraphDTA with 1D conv. (16) protein and GCN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.906 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepGraphDTA (1D conv. (16); GCN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepGraphDTA' with protein '1D conv. (16)' and drug 'GCN', column 'CI' (text-layer line 533)
Configuration: DeepGraphDTA with 1D conv. (16) protein and GCN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.174 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepGraphDTA (1D conv. (16); GCN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepGraphDTA' with protein '1D conv. (16)' and drug 'GCN', column 'MSE' (text-layer line 533)
Configuration: DeepGraphDTA with 1D conv. (32) protein and GIN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.909 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepGraphDTA (1D conv. (32); GIN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepGraphDTA' with protein '1D conv. (32)' and drug 'GIN', column 'CI' (text-layer line 536)
Configuration: DeepGraphDTA with 1D conv. (32) protein and GIN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.172 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepGraphDTA (1D conv. (32); GIN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepGraphDTA' with protein '1D conv. (32)' and drug 'GIN', column 'MSE' (text-layer line 536)
Configuration: DeepGraphDTA with 1D conv. (8) protein and GAT-GCN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.906 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepGraphDTA (1D conv. (8); GAT-GCN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepGraphDTA' with protein '1D conv. (8)' and drug 'GAT-GCN', column 'CI' (text-layer line 532)
Configuration: DeepGraphDTA with 1D conv. (8) protein and GAT-GCN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.175 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source DeepGraphDTA (1D conv. (8); GAT-GCN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'DeepGraphDTA' with protein '1D conv. (8)' and drug 'GAT-GCN', column 'MSE' (text-layer line 532)
Configuration: GraphDTA with 1D conv. protein and GAT drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.893 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source GraphDTA (1D conv.; GAT) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'GraphDTA' with protein '1D conv.' and drug 'GAT', column 'CI' (text-layer line 526)
Configuration: GraphDTA with 1D conv. protein and GAT drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.255 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source GraphDTA (1D conv.; GAT) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'GraphDTA' with protein '1D conv.' and drug 'GAT', column 'MSE' (text-layer line 526)
Configuration: GraphDTA with 1D conv. protein and GAT-GCN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.889 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source GraphDTA (1D conv.; GAT-GCN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'GraphDTA' with protein '1D conv.' and drug 'GAT-GCN', column 'CI' (text-layer line 525)
Configuration: GraphDTA with 1D conv. protein and GAT-GCN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.227 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source GraphDTA (1D conv.; GAT-GCN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'GraphDTA' with protein '1D conv.' and drug 'GAT-GCN', column 'MSE' (text-layer line 525)
Configuration: GraphDTA with 1D conv. protein and GCN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.88 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source GraphDTA (1D conv.; GCN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'GraphDTA' with protein '1D conv.' and drug 'GCN', column 'CI' (text-layer line 523)
Configuration: GraphDTA with 1D conv. protein and GCN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.237 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source GraphDTA (1D conv.; GCN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'GraphDTA' with protein '1D conv.' and drug 'GCN', column 'MSE' (text-layer line 523)
Configuration: GraphDTA with 1D conv. protein and GIN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.897 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source GraphDTA (1D conv.; GIN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'GraphDTA' with protein '1D conv.' and drug 'GIN', column 'CI' (text-layer line 527)
Configuration: GraphDTA with 1D conv. protein and GIN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.199 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source GraphDTA (1D conv.; GIN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'GraphDTA' with protein '1D conv.' and drug 'GIN', column 'MSE' (text-layer line 527)
Configuration: KronRLS with Similarity vector protein and Similarity vector drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.87 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source KronRLS (Similarity vector; Similarity vector) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'KronRLS' with protein 'Similarity vector' and drug 'Similarity vector', column 'CI' (text-layer line 522)
Configuration: KronRLS with Similarity vector protein and Similarity vector drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.359 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source KronRLS (Similarity vector; Similarity vector) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'KronRLS' with protein 'Similarity vector' and drug 'Similarity vector', column 'MSE' (text-layer line 522)
Configuration: PGraphDTA with ESM2 650M protein and GAT drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.898 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source PGraphDTA (ESM2 650M; GAT) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'PGraphDTA' with protein 'ESM2 650M' and drug 'GAT', column 'CI' (text-layer line 528)
Configuration: PGraphDTA with ESM2 650M protein and GAT drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.206 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source PGraphDTA (ESM2 650M; GAT) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'PGraphDTA' with protein 'ESM2 650M' and drug 'GAT', column 'MSE' (text-layer line 528)
Configuration: PLM-GraphDTA with ESMC 600M protein and GAT drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.908 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source PLM-GraphDTA (ESMC 600M; GAT) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'PLM-GraphDTA' with protein 'ESMC 600M' and drug 'GAT', column 'CI' (text-layer line 535)
Configuration: PLM-GraphDTA with ESMC 600M protein and GAT drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.174 MSE (Davis DeepDTA holdout fold) unit-unreported · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source PLM-GraphDTA (ESMC 600M; GAT) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'PLM-GraphDTA' with protein 'ESMC 600M' and drug 'GAT', column 'MSE' (text-layer line 535)
Configuration: PLM-GraphDTA with ESMC 600M protein and GAT-GCN drug representation	Protocol: Davis drug-target affinity, DeepDTA holdout fold Dataset: Davis holdout test fold (DeepDTA split)	0.91 Concordance index (Davis DeepDTA holdout fold) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Evidence concern: excluded from comparisons Table 3 KIBA rows KronRLS, SimBoost and DeepDTA equal the values printed in Ozturk et al. 2018, while the proposed models are 5-seed means; the rows may not share a training run or protocol. Do not compare these three rows against the proposed models automatically.Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 3, rows KronRLS, SimBoost, DeepDTA Methods, coverage and source PLM-GraphDTA (ESMC 600M; GAT-GCN) on Davis lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout Aggregation: Not reported Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Table 2, row 'PLM-GraphDTA' with protein 'ESMC 600M' and drug 'GAT-GCN', column 'CI' (text-layer line 538)

Source checking is not independent reproduction. Release 2026-10-22-1b7cf027e4f6.

Methods and evaluation design

Procedure, tasks and evaluated configurations

Recorded evaluations

Each evaluation records what was tested and under which conditions.

Explore all linked results

Baseline coverage

Reference methods help show what a model adds beyond simple controls. We track a null control and a conventional method for each protocol.

0 of 2 active baseline roles have published Rewire measurements in this release. Measurements on a selected protocol do not establish coverage of an entire suite.

No execution recipe linked to this protocol. Recipe availability does not establish a completed evaluation.

Author-reported evaluations: 8
External evaluations: 8

Literature evidence is not a Rewire measurement. Executed but unpublished runs and private review status are not included.

Null control

Proposed control: requires review

Training-set mean where supervised fitting is permitted

Protocol-specific applicability, permitted inputs, access, split, evaluator and execution requirements need review before implementation or execution.

This is a suggested selection rule, not a validated method or a measured score.

Conventional reference

Proposed control: requires review

Simple features with train-only ridge regression

Protocol-specific applicability, permitted inputs, access, split, evaluator and execution requirements need review before implementation or execution.

This is a suggested selection rule, not a validated method or a measured score.

Protocol coverage CSV (gzip) · Model evaluation matrix (gzip) · Source table (gzip) · Release and checksums (gzip)

Coverage is derived from release 2026-10-22-1b7cf027e4f6. Source citations describe the original records; they do not validate an unreviewed baseline proposal. No results have been generated by this audit.

Run instructions

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

Strengths, limitations and unresolved questions

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 2026-10-22-1b7cf027e4f6
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 2026-10-22-1b7cf027e4f6 · Record review: source checked

1 source record and release history

Comprehensive Evaluation of Protein Language Model Embeddings for Drug-Target Affinity Prediction · Original source · bioRxiv 10.64898/2026.08.31.748056v1 (posted 2026-09-03); 24-page PDF

Download this release (gzip)

Technical metadata and extraction receipts

Stable ID: lit-interactions-20261011-plmdta-protocol-davis-deepdta-holdout

areas: molecular-interactions
contexts: research
protocol: Davis DeepDTA holdout test fold; concordance index and MSE
version: DeepDTA split
limitations: Proposed models: mean over 5 random seeds (Section 3.4.3). Appendix A states that all models, baselines included, were trained and evaluated on the corrected Davis set; the seed averaging of baselines is not stated.; Rows are sorted by CI and only the selected PLM per drug backbone is shown.
source locator: Table 2

Related records

part of: Davis and KIBA drug-target affinity benchmark (DeepDTA split)
uses data: Davis holdout test fold (DeepDTA split)
assessment: DeepDTA (1D conv.; 1D conv.) on Davis
assessment: DeepGraphDTA (1D conv. (16); GAT) on Davis
assessment: DeepGraphDTA (1D conv. (16); GCN) on Davis
assessment: DeepGraphDTA (1D conv. (32); GIN) on Davis
assessment: DeepGraphDTA (1D conv. (8); GAT-GCN) on Davis
assessment: GraphDTA (1D conv.; GAT) on Davis
assessment: GraphDTA (1D conv.; GAT-GCN) on Davis
assessment: GraphDTA (1D conv.; GCN) on Davis
assessment: GraphDTA (1D conv.; GIN) on Davis
assessment: KronRLS (Similarity vector; Similarity vector) on Davis
assessment: PGraphDTA (ESM2 650M; GAT) on Davis
assessment: PLM-GraphDTA (ESMC 600M; GAT) on Davis
assessment: PLM-GraphDTA (ESMC 600M; GAT-GCN) on Davis
assessment: PLM-GraphDTA (ESMC 600M; GCN) on Davis
assessment: PLM-GraphDTA (ESMC 600M; GIN) on Davis
assessment: SimBoost (Similarity features; Similarity features) on Davis

Suggest a correction