rewire.itbenchmarks
Task

cell-type annotation

Cell-type annotation evaluates representations on datasets selected to be separate from pretraining data.

SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50

1 evaluation · 1 result

Overview

Metrics

Table 1 reports precision, recall and macro F1 separately for human immune cells and held-out non-immune cells. The table describes its ± terms as confidence intervals across three random initializations of test splits; the caption does not name a confidence level.

SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Table 1 caption and column headings
Evaluation procedure diagram
How it worksComputational evaluation flow
Computational evaluation flow1. Input: Single-cell expression representations.. Then: 2. Evaluation: The source states that evaluation datasets do not overlap with pretraining datasets.. Then: 3. Readout: Table 1 reports precision, recall and macro F1 separately for human immune cells and held-out non-immune cells. The table describes its ± terms as confidence intervals across three random initializations of test splits; the caption does not name a confidence level.Computational evaluation flow1. Input: Single-cell expression representations.. Then: 2. Evaluation: The source states that evaluation datasets do not overlap with pretraining datasets.. Then: 3. Readout: Table 1 reports precision, recall and macro F1 separately for human immune cells and held-out non-immune cells. The table describes its ± terms as confidence intervals across three random initializations of test splits; the caption does not name a confidence level.Computational evaluation flow1. Input: Single-cell expression representations.. Then: 2. Evaluation: The source states that evaluation datasets do not overlap with pretraining datasets.. Then: 3. Readout: Table 1 reports precision, recall and macro F1 separately for human immune cells and held-out non-immune cells. The table describes its ± terms as confidence intervals across three random initializations of test splits; the caption does not name a confidence level.

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

SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50; Table 1 caption and column headings

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

Results

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

All evaluations

1 evaluation · 1 result. Different protocols are not a single leaderboard.

Filter evaluations

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configurationProtocol and datasetFindingEvidence and details
Configuration: GREmLNTask: cell-type annotation
Dataset: non-immune cells
0.937 F1
fraction · unknown

Uncertainty: Not reported

Coverage: Not reported scored / Not reported eligible

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

GREmLN: cell-type annotation

Zero-shot cell-type annotation using pre-trained cellular graph foundation model

Aggregation: Not reported

GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Table 2, Cell type annotation(zero-shot), Non-immune cells, F1 row, GREmLN column

Source checking is not independent reproduction. Release 2026-09-29-06401fd5b220.

Methods and evaluation design

Procedure, tasks and evaluated configurations

How it works

Evaluation methodology

Healthy immune-cell data, cancer-infiltrating immune-cell data and non-immune-cell datasets. The source states that evaluation datasets do not overlap with pretraining datasets. Table 1 reports precision, recall and macro F1 separately for human immune cells and held-out non-immune cells. The table describes its ± terms as confidence intervals across three random initializations of test splits; the caption does not name a confidence level. The human immune-cell experiment compares GREmLN with scGPT, scFoundation and Geneformer. scGPT is excluded from the held-out non-immune-cell comparison because its CELLxGENE pretraining prevents that evaluation from being zero-shot. Test-cell regulatory graphs combine training-derived graphs using a training-fitted classifier rather than true test-cell labels.

SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50; Table 1 caption and column headings; Validation Experiments: Training cell type classifier; Table 1

Recorded evaluations

Each evaluation records what was tested and under which conditions.

Run instructions

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

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

Strengths, limitations and unresolved questions

Strengths and limitations

Strengths and considerations

No source-reviewed explanatory claims are recorded here yet.

Limitations and conditions

  • The immune-cell and held-out non-immune-cell settings have different comparator eligibility. scGPT’s absence from the latter is a pretraining-overlap control, not a failed prediction.
    SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Table 1 caption and column headings; Validation Experiments: Training cell type classifier; Table 1
Profile review details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Stable record: reported-task-6312c8a7ac045e

Specifications

Inputs, training, access and other details

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

Data, procedure and scoring
PropertyDescription and evidence
DatasetsHealthy immune-cell data, cancer-infiltrating immune-cell data and non-immune-cell datasets.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50
SplitsThe source states that evaluation datasets do not overlap with pretraining datasets.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50
MetricsTable 1 reports precision, recall and macro F1 separately for human immune cells and held-out non-immune cells. The table describes its ± terms as confidence intervals across three random initializations of test splits; the caption does not name a confidence level.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Table 1 caption and column headings
BaselinesThe human immune-cell experiment compares GREmLN with scGPT, scFoundation and Geneformer. scGPT is excluded from the held-out non-immune-cell comparison because its CELLxGENE pretraining prevents that evaluation from being zero-shot.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Validation Experiments: Training cell type classifier; Table 1
Leakage controlsTest-cell regulatory graphs combine training-derived graphs using a training-fitted classifier rather than true test-cell labels.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50
UncertaintyThe cited text-accessible evaluation sections give no confidence-interval, resampling or repeat-run error-bar specification. Image-only tables and uninspected supplements are outside this absence claim. · Not reported in inspected sources
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50
Entity typePaper-specific computational evaluation protocol.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50
OrganismsHuman immune cells and separately held-out non-immune cell collections.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50
AssaysSingle-cell expression with cell-type annotations.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50
Allowed inputsSingle-cell expression representations.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50
AdaptationZero-shot cell-type annotation is evaluated on held-out non-immune cells; fine-tuned perturbation-label prediction is a separate task.
SourcesGREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model · Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50

Evidence

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

Papers and result coverage

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

Paper or primary resourceVersionReference
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Modelpreprint version in PMCRead source
Historical gaps recorded on 2026-09-17

The catalogue now holds 1 result rows for this benchmark. A note below about pending extraction describes the state on 2026-09-17 and may since have been answered by a later batch. The result rows and their sources are the current record.

  • complete numerical transcription and independent cell review: Full primary artifact and table inventory preserved; no new numeric row is published from this audit alone.
  • exact checkpoint hashes and per-method scored denominators: Table labels alone do not establish these fields; do not infer checkpoint or scored count from model name or dataset size.
Search and extraction details

primary comparison tables located

Searches

  • GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model 10.1101/2025.07.03.663009

Evidence locations

  • Table 1:; XML table T1

Evidence table

Inspect claims, sources and review details

Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.

One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.

17 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-09-29-06401fd5b220
Property and statementOriginal source and locationReview and provenance
Diagram caption
Conceptual summary of the cited evaluation; exact task configuration and source version remain part of the protocol.
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50; Table 1 caption and column headings

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

source checked

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.diagram.caption

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Diagram steps
  • Input: Single-cell expression representations.
  • Evaluation: The source states that evaluation datasets do not overlap with pretraining datasets.
  • Readout: Table 1 reports precision, recall and macro F1 separately for human immune cells and held-out non-immune cells. The table describes its ± terms as confidence intervals across three random initializations of test splits; the caption does not name a confidence level.
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50; Table 1 caption and column headings

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

source checked

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.diagram.steps

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Diagram title
Computational evaluation flow
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50; Table 1 caption and column headings

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

source checked

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.diagram.title

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Datasets
Healthy immune-cell data, cancer-infiltrating immune-cell data and non-immune-cell datasets.
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

source checked

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.facts.0.value

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Splits
The source states that evaluation datasets do not overlap with pretraining datasets.
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

source checked

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.facts.1.value

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Adaptation
Zero-shot cell-type annotation is evaluated on held-out non-immune cells; fine-tuned perturbation-label prediction is a separate task.
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

source checked

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.facts.10.value

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Metrics
Table 1 reports precision, recall and macro F1 separately for human immune cells and held-out non-immune cells. The table describes its ± terms as confidence intervals across three random initializations of test splits; the caption does not name a confidence level.
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Table 1 caption and column headings

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

source checked

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.facts.2.value

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Baselines
The human immune-cell experiment compares GREmLN with scGPT, scFoundation and Geneformer. scGPT is excluded from the held-out non-immune-cell comparison because its CELLxGENE pretraining prevents that evaluation from being zero-shot.
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Validation Experiments: Training cell type classifier; Table 1

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

source checked

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.facts.3.value

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Leakage controls
Test-cell regulatory graphs combine training-derived graphs using a training-fitted classifier rather than true test-cell labels.
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

source checked

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.facts.4.value

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Uncertainty
The cited text-accessible evaluation sections give no confidence-interval, resampling or repeat-run error-bar specification. Image-only tables and uninspected supplements are outside this absence claim.
Individual claims
GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model

Original source ↗

Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50

Version: preprint version in PMC
Retrieved: 2026-09-16T10:33:57.502Z

unreported

automated source review · 2026-09-16

Audit details

Targeted full-paper and supplement review of the outstanding task fields, with original dataset metadata checked where accessible. Source-scoped omissions are explicit; no independent benchmark reproduction or numerical-result change.

Field: attributes.profile.facts.5.value

Source artifact SHA-256: 3a20c4ededb749fc3f1120baf16dcfebe3fcb30418a91c445cfd91a7b5fdf553

Hash scope: Hash scope not separately documented; inspect source record

Inspected artifact

Sources and history

View linked audit checks and correction history

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

2 source records and release historyDownload this release
Technical metadata and extraction receipts

Stable ID: reported-task-6312c8a7ac045e

areas
cells-tissues
tasks
cell-type annotation
entity level
task
version
Not reported
task
cell-type annotation
scope note
Paper-specific evaluation task; protocol completeness requires further extraction.
benchmark research
review date: 2026-09-17; status: primary_comparison_tables_located; primary sources: evidence-expansion-gremln-2026-3a20c4ed; inspected locators: Table 1:; XML table T1; searched queries: GREmLN: A Cellular Graph Structure Aware Transcriptomics Foundation Model 10.1101/2025.07.03.663009; gaps: complete numerical transcription and independent cell review: Full primary artifact and table inventory preserved; no new numeric row is published from this audit alone.; exact checkpoint hashes and per-method scored denominators: Table labels alone do not establish these fields; do not infer checkpoint or scored count from model name or dataset size.; claim scope: Dated primary-source discovery and protocol/table screening. Source checking does not mean experimental reproduction. Only separately extracted and independently reviewed numeric batches are publishable.
historical missing metadata
protocol version: not_reported_in_legacy_extract
metadata review scope
historical_missing_metadata preserves the original discovery state. Current descriptive evidence and missingness are recorded in profile.facts; numerical-result review is separate.
legacy kinds
benchmark
entity classification
review date: 2026-09-17; rationale: This source-scoped record identifies the biological prediction task and holds its paper context. Preserve the existing task identity; exact split, model adaptation and scoring remain in linked evaluations or separate protocol records.; source ids: gremln-2026; source locator: Evaluation Datasets; Transcriptomic Landscape Learning & Cell Type Annotation; Bayesian Graph Integration; cached text lines 39–50; ambiguities: A paper- or suite-specific task may constrain some inputs or metrics; that alone does not make it interchangeable with a complete versioned protocol. No protocol equivalence is inferred.; Some legacy profile Entity type facts use the generic phrase computational evaluation protocol. That boilerplate is not sufficient to establish a single fixed protocol identity or to merge this task with another protocol record.
Related records

Suggest a correction