Model

GlycanGT

GlycanGT learns glycan representations by treating monosaccharides and linkages as graph tokens.

Sources (5)

matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row

Results are available for configurations using this model. Their fitted heads, extra inputs and evaluation settings are kept separate below.

1 evaluated configuration using this model

How it worksGlycanGT workflow

Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation.

Sources (5)

Overview

Model type

Glycan graph transformer

Inputs

Glycan graphs with monosaccharide nodes and glycosidic-linkage edges.

Outputs

Graph embeddings and masked-component predictions for incomplete glycans.

Access

Official project documentation and implementation: https://github.com/matsui-lab/GlycanGT

Sources (5)

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

Related configurations, pipelines and services

These configurations, services and pipelines use this model within their own configurations. Their results, where available, are not assigned to the underlying model.

GlycanGT Pipeline · 21 results

Use this model

How it works, versions and access

How it works

GlycanGT turns each monosaccharide and glycosidic linkage into a token. Content embeddings, orthogonal node identifiers and token-type embeddings preserve graph structure before transformer attention. A graph token provides the whole-glycan representation; masked-token heads can suggest missing components, while downstream classifiers are fitted separately.

Sources (5)

Versions and reproducibility

ss, small, medium and large scales; the documented downstream work uses the large model with 35% masking. The applicable input limits require configuration-specific checking.

Sources (5)

Strengths, limitations and unresolved questions

Strengths and limitations

Strengths and considerations

Represents linkages explicitly and supports prediction of ambiguous glycan components.
Sources (5)
matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row

Limitations and conditions

The training corpus excluded ambiguous entries. Downstream taxonomy, glycosylation and immunogenicity tasks require their own evaluation data and heads.
Sources (5)
matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row

Profile review details

Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied.

Stable record: discovery-model-glycangt

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.

Inputs, outputs and configuration
Property	Description and evidence
Model type	Glycan graph transformer Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Architecture	TokenGT-derived graph transformer with content features, orthogonal node identifiers, node/edge type embeddings and a graph token. Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Inputs	Glycan graphs with monosaccharide nodes and glycosidic-linkage edges. Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Outputs	Graph embeddings and masked-component predictions for incomplete glycans. Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Parameters	ss: 1,922,002; small: 6,208,722; medium: 29,731,026; large: 91,820,242 total parameters, as printed in supplementary Table S2. Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Known versions	ss, small, medium and large scales; the documented downstream work uses the large model with 35% masking. Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Training data	The primary paper reports 83,739 curated glycans after ambiguity and downstream-overlap exclusions; the repository README reports 83,740. Both source values are preserved pending an author correction. Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Training cutoff	GlyCosmos/GlyTouCan source retrieval: 17 June 2025. The paper describes removal of downstream-benchmark overlaps from pretraining. Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Context limits	The inspected model presets set graph-identifier dimensions and attention sizes, not a validated maximum glycan node count. Token length depends on both monosaccharides and linkages. · Not reported in inspected sources Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Weights licence	Apache-2.0 declared in the author-linked Akikitani295/GlycanGT model-card metadata. Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Access	Official project documentation and implementation: https://github.com/matsui-lab/GlycanGT Sources (5) matsui-lab/GlycanGT: README.md; matsui-lab/GlycanGT: model/config_tokengt.py; Akikitani295/GlycanGT: README.md; glycangt: Journal full-text XML; glycangt-supp: Publisher supplementary archive · GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row
Code licence	Apache-2.0 Sources matsui-lab/GlycanGT: LICENSE · LICENSE: licence text

Applicable tests and references

Applicability is distinct from a completed evaluation.

GlycanML · Proposed association

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.

96 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 documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims	matsui-lab/GlycanGT: README.md Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 96611518c971deb89215ca163deaf9de3a59fa32 Retrieved: 2026-09-16T19:46:19.737544+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `fa39bffca31211baedb3b63df5dede775fcffff7b411d4261d5b18f526ae1153` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims	matsui-lab/GlycanGT: model/config_tokengt.py Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 96611518c971deb89215ca163deaf9de3a59fa32 Retrieved: 2026-09-16T19:46:19.737544+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `355c055f44e4b09c33ed640606800014b411e96674a1cdb48157c6d57ea5d426` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims	glycangt: Journal full-text XML Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Primary article XML snapshot Retrieved: 2026-09-16T20:20:56.439056+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `53e89a636c868c0329ee7eb6ae92f1028ec891940bc61730a148981b647fbbe5` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims	glycangt-supp: Publisher supplementary archive Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Retrieved page snapshot; no immutable publisher revision supplied Retrieved: 2026-09-16T20:42:10.495586+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `17987fb07979d88c8f073905247bdacd22e4607e16c3c399bb292685e711846f` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims	Akikitani295/GlycanGT: README.md Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 7c9f4a19b366ca48ed03995a0ad47fa1e745f24c Retrieved: 2026-09-16T20:04:02.231272+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `16a2967195f6be2d4478faf7151ceb7b96de4fa97ff168c7c975dc01d1dd69d3` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps Glycan graph Node and edge tokens Graph transformer Embedding or masked component Individual claims	matsui-lab/GlycanGT: README.md Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 96611518c971deb89215ca163deaf9de3a59fa32 Retrieved: 2026-09-16T19:46:19.737544+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `fa39bffca31211baedb3b63df5dede775fcffff7b411d4261d5b18f526ae1153` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps Glycan graph Node and edge tokens Graph transformer Embedding or masked component Individual claims	matsui-lab/GlycanGT: model/config_tokengt.py Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 96611518c971deb89215ca163deaf9de3a59fa32 Retrieved: 2026-09-16T19:46:19.737544+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `355c055f44e4b09c33ed640606800014b411e96674a1cdb48157c6d57ea5d426` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps Glycan graph Node and edge tokens Graph transformer Embedding or masked component Individual claims	glycangt: Journal full-text XML Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Primary article XML snapshot Retrieved: 2026-09-16T20:20:56.439056+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `53e89a636c868c0329ee7eb6ae92f1028ec891940bc61730a148981b647fbbe5` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps Glycan graph Node and edge tokens Graph transformer Embedding or masked component Individual claims	glycangt-supp: Publisher supplementary archive Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: Retrieved page snapshot; no immutable publisher revision supplied Retrieved: 2026-09-16T20:42:10.495586+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `17987fb07979d88c8f073905247bdacd22e4607e16c3c399bb292685e711846f` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact Inspected artifact
Diagram steps Glycan graph Node and edge tokens Graph transformer Embedding or masked component Individual claims	Akikitani295/GlycanGT: README.md Original source ↗ GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 7c9f4a19b366ca48ed03995a0ad47fa1e745f24c Retrieved: 2026-09-16T20:04:02.231272+00:00	source checked automated source review · 2026-09-16 Audit details Inspected pinned official documentation, relevant implementation files and named primary-paper sections. Claims are limited to those artifacts. Remaining field extraction and identity conflicts are explicit; no new performance claims, model runs or human review are implied. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `16a2967195f6be2d4478faf7151ceb7b96de4fa97ff168c7c975dc01d1dd69d3` Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact 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: discovered

7 source records and release history

matsui-lab/GlycanGT official source · Original source · 96611518c971deb89215ca163deaf9de3a59fa32
matsui-lab/GlycanGT: README.md · Original source · 96611518c971deb89215ca163deaf9de3a59fa32
matsui-lab/GlycanGT: model/config_tokengt.py · Original source · 96611518c971deb89215ca163deaf9de3a59fa32
Akikitani295/GlycanGT: README.md · Original source · 7c9f4a19b366ca48ed03995a0ad47fa1e745f24c
glycangt: Journal full-text XML · Original source · Primary article XML snapshot
glycangt-supp: Publisher supplementary archive · Original source · Retrieved page snapshot; no immutable publisher revision supplied
matsui-lab/GlycanGT: LICENSE · Original source · 96611518c971deb89215ca163deaf9de3a59fa32

Download this release

Technical metadata and extraction receipts

Stable ID: discovery-model-glycangt

areas: glycomics
access: official_source_linked
benchmark applicability: candidate; not evidence of a reported evaluation
candidate benchmark ids: discovery-benchmark-glycanml
entity level: family
reported name: GlycanGT
version: Not reported
historical missing metadata: checkpoint: unextracted; code licence: unextracted; parameters: unextracted; training cutoff: unextracted; training data: unextracted; version: unextracted; weights licence: unextracted
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.
entity classification: review date: 2026-09-17; rationale: The cited profile describes a named learned biological predictor or representation model/family. Preserve this identity separately from task-specific fitting, individual checkpoints, pipelines and hosted access.; source ids: evidence-official-10a6550ca70fdeaf33cb; evidence-official-790ab6283e5bb91474ed; evidence-official-dacb012b22c0a7de4880; evidence-official-87c70bddba97a72a5e5d; evidence-official-ba12a2d759aba6800ae6; source locator: GlycanGT paper Sections 2.1–2.4,2.9 and 4; official model-card licence metadata; primary paper Sections 2.1 and 3.1 versus README Model architecture / Training details; model/config_tokengt.py: size presets; publisher supplementary Table S2, Total Parameters row; ambiguities: None recorded

Related records

applicable to: GlycanML
uses model: GlycanGT

Suggest a correction