rewirebio.iobenchmarks
Configuration

GlycanAA

GlycanAA hierarchical relational graph baseline with atom and monosaccharide nodes, trained and evaluated on the same GlycanML task splits; exact trained checkpoint unreported.

20 evaluations · 20 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.

Model Capabilities & Hardware Requirements

Deployment facts recorded for GlycanAA. Fields the record does not carry are marked not recorded.

ParametersNot recorded
Minimum VRAMNot recorded
Inference LatencyNot recorded
Context WindowNot recorded
AccessNot recorded
LicenseNot recorded

Evaluations and results

20 evaluations · 20 results. Different protocols are not a single leaderboard.

Sorted by Accuracy (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 configurationProtocol and datasetAccuracyMacro F1AUPRC
GlycanAAEvaluationGlycanML glycosylation Accuracy: GlycanGT study: glycosylation Accuracy, GlyConnect glycosylation; GlycanML official motif split (GlycanML split)0.96950000000000003 ± 0 (best in column)Not reportedNot reported
GlycanAAEvaluationGlycanML domain Accuracy: GlycanGT study: domain Accuracy, SugarBase taxonomy domain; GlycanML official motif split (GlycanML split)0.93473333333333297 ± 0.0117457793838184Not reportedNot reported
GlycanAAEvaluationGlycanML kingdom Accuracy: GlycanGT study: kingdom Accuracy, SugarBase taxonomy kingdom; GlycanML official motif split (GlycanML split)0.92420000000000002 ± 0.015649920127591699Not reportedNot reported
GlycanAAEvaluationGlycanML phylum Accuracy: GlycanGT study: phylum Accuracy, SugarBase taxonomy phylum; GlycanML official motif split (GlycanML split)0.86686666666666601 ± 0.0142043420591498Not reportedNot reported
GlycanAAEvaluationGlycanML immunogenicity Accuracy: GlycanGT study: immunogenicity Accuracy, SugarBase immunogenicity; GlycanML official motif split (GlycanML split)0.86209999999999998 ± 0Not reportedNot reported
GlycanAAEvaluationGlycanML class Accuracy: GlycanGT study: class Accuracy, SugarBase taxonomy class; GlycanML official motif split (GlycanML split)0.74646666666666595 ± 0.0136313364470742Not reportedNot reported
GlycanAAEvaluationGlycanML order Accuracy: GlycanGT study: order Accuracy, SugarBase taxonomy order; GlycanML official motif split (GlycanML split)0.49836666666666601 ± 0.0066335008354060998Not reportedNot reported
GlycanAAEvaluationGlycanML family Accuracy: GlycanGT study: family Accuracy, SugarBase taxonomy family; GlycanML official motif split (GlycanML split)0.43126666666666602 ± 0.086080911550315997Not reportedNot reported
GlycanAAEvaluationGlycanML species Accuracy: GlycanGT study: species Accuracy, SugarBase taxonomy species; GlycanML official motif split (GlycanML split)0.40549999999999897 ± 0.0054836119483420804Not reportedNot reported
GlycanAAEvaluationGlycanML genus Accuracy: GlycanGT study: genus Accuracy, SugarBase taxonomy genus; GlycanML official motif split (GlycanML split)0.38300000000000001 ± 0.0324371700368574Not reportedNot reported
GlycanAAEvaluationGlycanML class Macro-F1: GlycanGT study: class Macro-F1, SugarBase taxonomy class; GlycanML official motif split (GlycanML split)Not reported0.41666666666666602 ± 0.0177798575172393Not reported
GlycanAAEvaluationGlycanML domain Macro-F1: GlycanGT study: domain Macro-F1, SugarBase taxonomy domain; GlycanML official motif split (GlycanML split)Not reported0.62916666666666599 ± 0.011393126582871499Not reported
GlycanAAEvaluationGlycanML family Macro-F1: GlycanGT study: family Macro-F1, SugarBase taxonomy family; GlycanML official motif split (GlycanML split)Not reported0.212166666666666 ± 0.015383215961993499Not reported
GlycanAAEvaluationGlycanML genus Macro-F1: GlycanGT study: genus Macro-F1, SugarBase taxonomy genus; GlycanML official motif split (GlycanML split)Not reported0.18640000000000001 ± 0.0119478031453485Not reported
GlycanAAEvaluationGlycanML glycosylation Macro-F1: GlycanGT study: glycosylation Macro-F1, GlyConnect glycosylation; GlycanML official motif split (GlycanML split)Not reported0.95226666666666604 ± 0.0018475208614067899 (best in column)Not reported
GlycanAAEvaluationGlycanML immunogenicity AUPRC: GlycanGT study: immunogenicity AUPRC, SugarBase immunogenicity; GlycanML official motif split (GlycanML split)Not reportedNot reported0.70573333333333299 ± 0.069754736995657304
GlycanAAEvaluationGlycanML kingdom Macro-F1: GlycanGT study: kingdom Macro-F1, SugarBase taxonomy kingdom; GlycanML official motif split (GlycanML split)Not reported0.59389999999999998 ± 0.095006157695172497Not reported
GlycanAAEvaluationGlycanML order Macro-F1: GlycanGT study: order Macro-F1, SugarBase taxonomy order; GlycanML official motif split (GlycanML split)Not reported0.2727 ± 0.0137720731917892Not reported
GlycanAAEvaluationGlycanML phylum Macro-F1: GlycanGT study: phylum Macro-F1, SugarBase taxonomy phylum; GlycanML official motif split (GlycanML split)Not reported0.440066666666666 ± 0.0041428653530296202Not reported
GlycanAAEvaluationGlycanML species Macro-F1: GlycanGT study: species Macro-F1, SugarBase taxonomy species; GlycanML official motif split (GlycanML split)Not reported0.15840000000000001 ± 0.012304064369142401Not reported
All 20 result rows with coverage, uncertainty and sources
Filter evaluations

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configurationProtocol and datasetFindingEvidence and details
Configuration: GlycanAAProtocol: GlycanML class Accuracy: GlycanGT study: class Accuracy
Dataset subset: SugarBase taxonomy class; GlycanML official motif split (GlycanML split)
0.74646666666666595 ± 0.0136313364470742 Accuracy
fraction · higher

Uncertainty: SD 0.0136313364470742

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML class Accuracy: GlycanGT study: class Accuracy

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A24:E24 (mean D24, SD E24)
Configuration: GlycanAAProtocol: GlycanML class Macro-F1: GlycanGT study: class Macro-F1
Dataset subset: SugarBase taxonomy class; GlycanML official motif split (GlycanML split)
0.41666666666666602 ± 0.0177798575172393 Macro F1
unitless · higher

Uncertainty: SD 0.0177798575172393

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML class Macro-F1: GlycanGT study: class Macro-F1

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A25:E25 (mean D25, SD E25)
Configuration: GlycanAAProtocol: GlycanML domain Accuracy: GlycanGT study: domain Accuracy
Dataset subset: SugarBase taxonomy domain; GlycanML official motif split (GlycanML split)
0.93473333333333297 ± 0.0117457793838184 Accuracy
fraction · higher

Uncertainty: SD 0.0117457793838184

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML domain Accuracy: GlycanGT study: domain Accuracy

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A34:E34 (mean D34, SD E34)
Configuration: GlycanAAProtocol: GlycanML domain Macro-F1: GlycanGT study: domain Macro-F1
Dataset subset: SugarBase taxonomy domain; GlycanML official motif split (GlycanML split)
0.62916666666666599 ± 0.011393126582871499 Macro F1
unitless · higher

Uncertainty: SD 0.011393126582871499

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML domain Macro-F1: GlycanGT study: domain Macro-F1

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A35:E35 (mean D35, SD E35)
Configuration: GlycanAAProtocol: GlycanML family Accuracy: GlycanGT study: family Accuracy
Dataset subset: SugarBase taxonomy family; GlycanML official motif split (GlycanML split)
0.43126666666666602 ± 0.086080911550315997 Accuracy
fraction · higher

Uncertainty: SD 0.086080911550315997

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML family Accuracy: GlycanGT study: family Accuracy

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A26:E26 (mean D26, SD E26)
Configuration: GlycanAAProtocol: GlycanML family Macro-F1: GlycanGT study: family Macro-F1
Dataset subset: SugarBase taxonomy family; GlycanML official motif split (GlycanML split)
0.212166666666666 ± 0.015383215961993499 Macro F1
unitless · higher

Uncertainty: SD 0.015383215961993499

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML family Macro-F1: GlycanGT study: family Macro-F1

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A27:E27 (mean D27, SD E27)
Configuration: GlycanAAProtocol: GlycanML genus Accuracy: GlycanGT study: genus Accuracy
Dataset subset: SugarBase taxonomy genus; GlycanML official motif split (GlycanML split)
0.38300000000000001 ± 0.0324371700368574 Accuracy
fraction · higher

Uncertainty: SD 0.0324371700368574

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML genus Accuracy: GlycanGT study: genus Accuracy

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A36:E36 (mean D36, SD E36)
Configuration: GlycanAAProtocol: GlycanML genus Macro-F1: GlycanGT study: genus Macro-F1
Dataset subset: SugarBase taxonomy genus; GlycanML official motif split (GlycanML split)
0.18640000000000001 ± 0.0119478031453485 Macro F1
unitless · higher

Uncertainty: SD 0.0119478031453485

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML genus Macro-F1: GlycanGT study: genus Macro-F1

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A37:E37 (mean D37, SD E37)
Configuration: GlycanAAProtocol: GlycanML glycosylation Accuracy: GlycanGT study: glycosylation Accuracy
Dataset subset: GlyConnect glycosylation; GlycanML official motif split (GlycanML split)
0.96950000000000003 ± 0 Accuracy
fraction · higher

Uncertainty: SD 0

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML glycosylation Accuracy: GlycanGT study: glycosylation Accuracy

Glycosylation: 1,683 glycans total; N-linked/O-linked/free. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A38:E38 (mean D38, SD E38)
Configuration: GlycanAAProtocol: GlycanML glycosylation Macro-F1: GlycanGT study: glycosylation Macro-F1
Dataset subset: GlyConnect glycosylation; GlycanML official motif split (GlycanML split)
0.95226666666666604 ± 0.0018475208614067899 Macro F1
unitless · higher

Uncertainty: SD 0.0018475208614067899

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML glycosylation Macro-F1: GlycanGT study: glycosylation Macro-F1

Glycosylation: 1,683 glycans total; N-linked/O-linked/free. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A39:E39 (mean D39, SD E39)
Configuration: GlycanAAProtocol: GlycanML immunogenicity Accuracy: GlycanGT study: immunogenicity Accuracy
Dataset subset: SugarBase immunogenicity; GlycanML official motif split (GlycanML split)
0.86209999999999998 ± 0 Accuracy
fraction · higher

Uncertainty: SD 0

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML immunogenicity Accuracy: GlycanGT study: immunogenicity Accuracy

Immunogenicity: 1,320 glycans total; binary immune activity. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A42:E42 (mean D42, SD E42)
Configuration: GlycanAAProtocol: GlycanML immunogenicity AUPRC: GlycanGT study: immunogenicity AUPRC
Dataset subset: SugarBase immunogenicity; GlycanML official motif split (GlycanML split)
0.70573333333333299 ± 0.069754736995657304 AUPRC
unitless · higher

Uncertainty: SD 0.069754736995657304

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML immunogenicity AUPRC: GlycanGT study: immunogenicity AUPRC

Immunogenicity: 1,320 glycans total; binary immune activity. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A43:E43 (mean D43, SD E43)
Configuration: GlycanAAProtocol: GlycanML kingdom Accuracy: GlycanGT study: kingdom Accuracy
Dataset subset: SugarBase taxonomy kingdom; GlycanML official motif split (GlycanML split)
0.92420000000000002 ± 0.015649920127591699 Accuracy
fraction · higher

Uncertainty: SD 0.015649920127591699

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML kingdom Accuracy: GlycanGT study: kingdom Accuracy

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A28:E28 (mean D28, SD E28)
Configuration: GlycanAAProtocol: GlycanML kingdom Macro-F1: GlycanGT study: kingdom Macro-F1
Dataset subset: SugarBase taxonomy kingdom; GlycanML official motif split (GlycanML split)
0.59389999999999998 ± 0.095006157695172497 Macro F1
unitless · higher

Uncertainty: SD 0.095006157695172497

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML kingdom Macro-F1: GlycanGT study: kingdom Macro-F1

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A29:E29 (mean D29, SD E29)
Configuration: GlycanAAProtocol: GlycanML order Accuracy: GlycanGT study: order Accuracy
Dataset subset: SugarBase taxonomy order; GlycanML official motif split (GlycanML split)
0.49836666666666601 ± 0.0066335008354060998 Accuracy
fraction · higher

Uncertainty: SD 0.0066335008354060998

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML order Accuracy: GlycanGT study: order Accuracy

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A30:E30 (mean D30, SD E30)
Configuration: GlycanAAProtocol: GlycanML order Macro-F1: GlycanGT study: order Macro-F1
Dataset subset: SugarBase taxonomy order; GlycanML official motif split (GlycanML split)
0.2727 ± 0.0137720731917892 Macro F1
unitless · higher

Uncertainty: SD 0.0137720731917892

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML order Macro-F1: GlycanGT study: order Macro-F1

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A31:E31 (mean D31, SD E31)
Configuration: GlycanAAProtocol: GlycanML phylum Accuracy: GlycanGT study: phylum Accuracy
Dataset subset: SugarBase taxonomy phylum; GlycanML official motif split (GlycanML split)
0.86686666666666601 ± 0.0142043420591498 Accuracy
fraction · higher

Uncertainty: SD 0.0142043420591498

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML phylum Accuracy: GlycanGT study: phylum Accuracy

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A40:E40 (mean D40, SD E40)
Configuration: GlycanAAProtocol: GlycanML phylum Macro-F1: GlycanGT study: phylum Macro-F1
Dataset subset: SugarBase taxonomy phylum; GlycanML official motif split (GlycanML split)
0.440066666666666 ± 0.0041428653530296202 Macro F1
unitless · higher

Uncertainty: SD 0.0041428653530296202

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML phylum Macro-F1: GlycanGT study: phylum Macro-F1

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A41:E41 (mean D41, SD E41)
Configuration: GlycanAAProtocol: GlycanML species Accuracy: GlycanGT study: species Accuracy
Dataset subset: SugarBase taxonomy species; GlycanML official motif split (GlycanML split)
0.40549999999999897 ± 0.0054836119483420804 Accuracy
fraction · higher

Uncertainty: SD 0.0054836119483420804

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML species Accuracy: GlycanGT study: species Accuracy

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A32:E32 (mean D32, SD E32)
Configuration: GlycanAAProtocol: GlycanML species Macro-F1: GlycanGT study: species Macro-F1
Dataset subset: SugarBase taxonomy species; GlycanML official motif split (GlycanML split)
0.15840000000000001 ± 0.012304064369142401 Macro F1
unitless · higher

Uncertainty: SD 0.012304064369142401

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

GlycanAA on GlycanML species Macro-F1: GlycanGT study: species Macro-F1

Taxonomy: 13,209 glycans total across eight levels, 4–1,737 classes per level. Official fixed GlycanML motif-based train/validation/test splits (8:1:1). Section 2.5 reports class-balanced classifiers, train ∪ validation hyperparameter selection by randomized search with 3-fold cross-validation, followed by one evaluation on the held-out test set; complete procedure repeated with three random seeds, reporting mean and standard deviation. GlycanGT large model pretrained with 35% masking provides [Graph] embeddings to SVM/LightGBM; the selected classifier for each S4 row is not identified. Section 2.6 states that all four graph baselines were trained and evaluated on the same datasets/splits; it does not establish that each baseline used the GlycanGT downstream classifier search. SVM search: 10 iterations, RBF/linear, C logU(1e-3,1e2), gamma logU(1e-4,1e-1); LightGBM: 15 randomized iterations. Do not combine with separate original GlycanML-paper protocols.

Aggregation: Not reported

GlycanGT published supplementary archive, Table S4; glycangt: Journal full-text XML · btag147_supplementary_data.zip / Table_S4.xlsx / Sheet1!A33:E33 (mean D33, SD E33)

Source checking is not independent reproduction. Release 2026-11-21-c8bea25e5abb.

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 2026-11-21-c8bea25e5abb
Property and statementOriginal source and locationReview and provenance

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

Sources and history

Release 2026-11-21-c8bea25e5abb · Record review: source checked

2 source records and release historyDownload this release (gzip)
Technical metadata and record identifiers

Technical metadata for this record is tracked under stable ID glycangt-2026-table-s4-method-glycanaa.

Source locator: GlycanGT primary article Sections 2.1, 2.5, 2.6 and 3.1–3.2 (PMC13105845), Supplementary Table S4; column B GlycanAA

Related records

How to cite

GlycanAA. Rewire biological model benchmark database, release 2026-11-21-c8bea25e5abb. https://benchmarks.rewirebio.io/database/configuration/glycangt-2026-table-s4-method-glycanaa/

Release 2026-11-21-c8bea25e5abb. Rankings and values change between releases, so keep the release id in the citation. Cite the original publication as well for any value you reuse: each result links its source.

The compilation and annotations made by Rewire in this release are offered under CC BY 4.0; values and text taken from cited sources stay under the terms of those sources and are not covered by this licence. Data licence

Suggest a correction