Evaluation

CellPLM on Chen et al. 2026 genetic perturbation prediction, Norman

Published comparison transcribed from the pinned source; not reproduced.

Evaluation results

1 evaluation · 6 results. Different protocols are not a single leaderboard.

Protocol: Chen et al. 2026 genetic perturbation prediction, Norman · Dataset subset: Norman (Chen et al. 2026)

Sorted by MSE (all genes, predicted versus observed post-perturbation expression) (lower is better). The best value in each column is highlighted. Decimals are rounded for display; each value links to the printed value and its source.

Tested configuration	MSE (all genes, predicted versus observed post-perturbation expression)	MSE (top 20 differentially expressed genes per perturbation condition, predicted versus observed post-perturbation expression)	Pearson (all genes, predicted versus observed post-perturbation expression)	Pearson (top 20 differentially expressed genes per perturbation condition, predicted versus observed post-perturbation expression)	Pearson delta (all genes in the intersected gene set)	Pearson delta (top 20 differentially expressed genes per perturbation condition)
CellPLM (gears-decoder, Chen et al. 2026)Evaluation	0.0064	0.403	0.983	0.872	0.473	0.398

All 6 result rows with coverage, uncertainty and sources

Filter evaluations

Tested model or methodProtocolDatasetEvaluation setupMetricEvidence origin

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configuration	Protocol and dataset	Finding	Evidence and details
Configuration: CellPLM (gears-decoder, Chen et al. 2026)	Protocol: Chen et al. 2026 genetic perturbation prediction, Norman Dataset subset: Norman (Chen et al. 2026)	0.0064 MSE (all genes, predicted versus observed post-perturbation expression) unitless · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 4 precision and recall equal accuracy in all eight Nicheformer and CellPLM rows, as micro (global) averaging would give, but recall differs from accuracy in every scGPT-spatial, GenePT-w, scELMo and Novae row. The Methods call precision and recall 'global' without defining the averaging, so they may be averaged differently between models and should not be compared across models.Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 4, columns 'Accuracy', 'Precision' and 'Recall'; Methods 'Cell type annotation' (page 25) Methods, coverage and source CellPLM on Chen et al. 2026 genetic perturbation prediction, Norman lit-cells-networks-20261011-chen2026-protocol-perturbation-norman Aggregation: Not reported Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 5, model 'CellPLM', dataset 'Norman', column 'MSE'
Configuration: CellPLM (gears-decoder, Chen et al. 2026)	Protocol: Chen et al. 2026 genetic perturbation prediction, Norman Dataset subset: Norman (Chen et al. 2026)	0.403 MSE (top 20 differentially expressed genes per perturbation condition, predicted versus observed post-perturbation expression) unitless · lower Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 4 precision and recall equal accuracy in all eight Nicheformer and CellPLM rows, as micro (global) averaging would give, but recall differs from accuracy in every scGPT-spatial, GenePT-w, scELMo and Novae row. The Methods call precision and recall 'global' without defining the averaging, so they may be averaged differently between models and should not be compared across models.Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 4, columns 'Accuracy', 'Precision' and 'Recall'; Methods 'Cell type annotation' (page 25) Methods, coverage and source CellPLM on Chen et al. 2026 genetic perturbation prediction, Norman lit-cells-networks-20261011-chen2026-protocol-perturbation-norman Aggregation: Not reported Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 5, model 'CellPLM', dataset 'Norman', column 'MSE DE'
Configuration: CellPLM (gears-decoder, Chen et al. 2026)	Protocol: Chen et al. 2026 genetic perturbation prediction, Norman Dataset subset: Norman (Chen et al. 2026)	0.983 Pearson (all genes, predicted versus observed post-perturbation expression) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 4 precision and recall equal accuracy in all eight Nicheformer and CellPLM rows, as micro (global) averaging would give, but recall differs from accuracy in every scGPT-spatial, GenePT-w, scELMo and Novae row. The Methods call precision and recall 'global' without defining the averaging, so they may be averaged differently between models and should not be compared across models.Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 4, columns 'Accuracy', 'Precision' and 'Recall'; Methods 'Cell type annotation' (page 25) Methods, coverage and source CellPLM on Chen et al. 2026 genetic perturbation prediction, Norman lit-cells-networks-20261011-chen2026-protocol-perturbation-norman Aggregation: Not reported Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 5, model 'CellPLM', dataset 'Norman', column 'Pearson'
Configuration: CellPLM (gears-decoder, Chen et al. 2026)	Protocol: Chen et al. 2026 genetic perturbation prediction, Norman Dataset subset: Norman (Chen et al. 2026)	0.872 Pearson (top 20 differentially expressed genes per perturbation condition, predicted versus observed post-perturbation expression) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 4 precision and recall equal accuracy in all eight Nicheformer and CellPLM rows, as micro (global) averaging would give, but recall differs from accuracy in every scGPT-spatial, GenePT-w, scELMo and Novae row. The Methods call precision and recall 'global' without defining the averaging, so they may be averaged differently between models and should not be compared across models.Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 4, columns 'Accuracy', 'Precision' and 'Recall'; Methods 'Cell type annotation' (page 25) Methods, coverage and source CellPLM on Chen et al. 2026 genetic perturbation prediction, Norman lit-cells-networks-20261011-chen2026-protocol-perturbation-norman Aggregation: Not reported Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 5, model 'CellPLM', dataset 'Norman', column 'Pearson DE'
Configuration: CellPLM (gears-decoder, Chen et al. 2026)	Protocol: Chen et al. 2026 genetic perturbation prediction, Norman Dataset subset: Norman (Chen et al. 2026)	0.473 Pearson delta (all genes in the intersected gene set) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 4 precision and recall equal accuracy in all eight Nicheformer and CellPLM rows, as micro (global) averaging would give, but recall differs from accuracy in every scGPT-spatial, GenePT-w, scELMo and Novae row. The Methods call precision and recall 'global' without defining the averaging, so they may be averaged differently between models and should not be compared across models.Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 4, columns 'Accuracy', 'Precision' and 'Recall'; Methods 'Cell type annotation' (page 25) Methods, coverage and source CellPLM on Chen et al. 2026 genetic perturbation prediction, Norman lit-cells-networks-20261011-chen2026-protocol-perturbation-norman Aggregation: Not reported Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 5, model 'CellPLM', dataset 'Norman', column 'Pearson Delta'
Configuration: CellPLM (gears-decoder, Chen et al. 2026)	Protocol: Chen et al. 2026 genetic perturbation prediction, Norman Dataset subset: Norman (Chen et al. 2026)	0.398 Pearson delta (top 20 differentially expressed genes per perturbation condition) unitless · higher Uncertainty: Not reported by the source Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Evidence concern: excluded from comparisons Table 4 precision and recall equal accuracy in all eight Nicheformer and CellPLM rows, as micro (global) averaging would give, but recall differs from accuracy in every scGPT-spatial, GenePT-w, scELMo and Novae row. The Methods call precision and recall 'global' without defining the averaging, so they may be averaged differently between models and should not be compared across models.Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 4, columns 'Accuracy', 'Precision' and 'Recall'; Methods 'Cell type annotation' (page 25) Methods, coverage and source CellPLM on Chen et al. 2026 genetic perturbation prediction, Norman lit-cells-networks-20261011-chen2026-protocol-perturbation-norman Aggregation: Not reported Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Table 5, model 'CellPLM', dataset 'Norman', column 'Pearson Delta DE'

Source checking is not independent reproduction. Release 2026-10-23-5347695500ed.

Research readiness

0 of 4 readiness checks met. These checks assess whether the evidence supports a reproducible investigation; a source-checked score alone does not meet them.

Readiness checks, gaps and artifacts

Release 2026-10-23-5347695500ed · Evidence verified: Not verified

Evidence incomplete

Replay metrics

Exact outcomes, predictions, identifiers and evaluator are connected.

Missing or unresolved evidence

No verified artifact manifest is linked to this exact record.
File checksums match the recorded files: not yet verified
Predictions are matched to the right samples: not yet verified
Score meaning and direction are confirmed: not yet verified
Metrics are recomputed from the saved predictions: not yet verified

Verified: Not verified

Evidence incomplete

Investigate discrepancies

Replay evidence includes annotations and an assessment of dependence. Unknown independence permits descriptive analysis only.

Missing or unresolved evidence

No verified artifact manifest is linked to this exact record.
File checksums match the recorded files: not yet verified
Predictions are matched to the right samples: not yet verified
Score meaning and direction are confirmed: not yet verified
Metrics are recomputed from the saved predictions: not yet verified
Sample annotations are recorded: not yet verified
Dependence between samples is assessed: not yet verified

Verified: Not verified

Evidence incomplete

Run locally

A pinned recipe describes the inputs, environment and resource requirements.

Missing or unresolved evidence

No verified artifact manifest is linked to this exact record.
File checksums match the recorded files: not yet verified
Predictions are matched to the right samples: not yet verified
Score meaning and direction are confirmed: not yet verified
A pinned run recipe exists: not yet verified
Compute requirements are estimated: not yet verified

Verified: Not verified

Evidence incomplete

Validate independently

Separate data and exposure records support an independent test.

Missing or unresolved evidence

No verified artifact manifest is linked to this exact record.
File checksums match the recorded files: not yet verified
Predictions are matched to the right samples: not yet verified
Score meaning and direction are confirmed: not yet verified
Independent validation data exist: not yet verified
Overlap with training data is checked: not yet verified

Verified: Not verified

Readiness describes the evidence in this release. Availability on your computer is checked separately when an investigation runs. Existing data exposure can prevent independent validation even when files are available.

Artifacts and reproduction

No verified artifact manifest is connected to this record yet. The gaps above identify what is needed before analysis can begin.

Evaluation procedure

Chen et al. 2026 genetic perturbation prediction, Norman

Configuration: CellPLM (gears-decoder, Chen et al. 2026)
Protocol: Chen et al. 2026 genetic perturbation prediction, Norman
Dataset subset: Norman (Chen et al. 2026)

origin: Independent external evaluation
configuration: Primary source as retrieved 2026-10-11
protocol id: lit-cells-networks-20261011-chen2026-protocol-perturbation-norman
dataset version: Not reported
split: Perturbation split of the scGPT-style workflow
population: Not reported
inputs: Single-cell or spatial expression matrix, aligned to each model's gene vocabulary
adaptation: model embeddings (or gene embeddings for GenePT-w and scELMo) used with a shared GEARS perturbation decoder in the scGPT-style perturbation workflow
metric implementation: Not reported
aggregation: Not reported
budget: Not reported

Metadata review: source checked. Unreported conditions prevent automatic comparisons.

Reproduction

Split: Perturbation split of the scGPT-style workflow
Adaptation: model embeddings (or gene embeddings for GenePT-w and scELMo) used with a shared GEARS perturbation decoder in the scGPT-style perturbation workflow
Scoring implementation: Not reported

No execution recipe has been verified for this exact configuration and evaluation. A benchmark's general instructions may use different inputs, splits or model settings.

Reproducing this published result requires matching its model configuration, data, split and scorer. Source checking or a successful smoke test does not establish score reproduction.

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.

18 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-10-23-5347695500ed
Property and statement	Original source and location	Review and provenance
Comparison: adaptation model embeddings (or gene embeddings for GenePT-w and scELMo) used with a shared GEARS perturbation decoder in the scGPT-style perturbation workflow Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	not individually reviewed No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.comparison.adaptation` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Comparison: aggregation Not reported Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	missing or unspecified No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.comparison.aggregation` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Comparison: budget Not reported Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	missing or unspecified No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.comparison.budget` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Comparison: dataset version Not reported Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	missing or unspecified No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.comparison.dataset_version` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Comparison: inputs Single-cell or spatial expression matrix, aligned to each model's gene vocabulary Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	not individually reviewed No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.comparison.inputs` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Comparison: metric implementation Not reported Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	missing or unspecified No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.comparison.metric_implementation` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Comparison: population Not reported Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	missing or unspecified No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.comparison.population` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Comparison: protocol id lit-cells-networks-20261011-chen2026-protocol-perturbation-norman Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	not individually reviewed No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.comparison.protocol_id` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Comparison: split Perturbation split of the scGPT-style workflow Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	not individually reviewed No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.comparison.split` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Origin independent_paper Context-only references	Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation Original source ↗ Table 5, model 'CellPLM', dataset 'Norman' Version: arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19) Retrieved: 2026-10-11T21:29:34Z	not individually reviewed No individual claim review recorded independent paper Source has a recorded evidence concern. Consult its source page before using the claim. Audit details Field: `attributes.origin` Source artifact SHA-256: `8137338c6146f570a6546f30707d4244baca9dab848b32c51539665a5ba539da` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact

How to read the evidence

Sources and history

Release 2026-10-23-5347695500ed · Record review: source checked

1 source record and release history

Harmonised benchmarking of foundation models for single-cell and spatial transcriptomics reveals context-dependent generalisation · Original source · arXiv 2607.17227v2, updated 2026-09-17 (v1 posted 2026-07-19)

Download this release (gzip)

Technical metadata and extraction receipts

Stable ID: lit-cells-networks-20261011-chen2026-eval-perturbation-norman-cellplm

areas: single-cell; spatial-omics
contexts: research
origin: independent_paper
protocol: lit-cells-networks-20261011-chen2026-protocol-perturbation-norman
version: Primary source as retrieved 2026-10-11
comparison: protocol id: lit-cells-networks-20261011-chen2026-protocol-perturbation-norman; dataset version: Not reported; split: Perturbation split of the scGPT-style workflow; population: Not reported; inputs: Single-cell or spatial expression matrix, aligned to each model's gene vocabulary; adaptation: model embeddings (or gene embeddings for GenePT-w and scELMo) used with a shared GEARS perturbation decoder in the scGPT-style perturbation workflow; metric implementation: Not reported; aggregation: Not reported; budget: Not reported
source locator: Table 5, model 'CellPLM', dataset 'Norman'
missing metadata: comparison.aggregation: reason: unreported; comparison.budget: reason: unreported; comparison.dataset version: reason: unreported; comparison.metric implementation: reason: unreported; comparison.population: reason: unreported

Related records

system: CellPLM (gears-decoder, Chen et al. 2026)
assessment: Chen et al. 2026 genetic perturbation prediction, Norman
data: Norman (Chen et al. 2026)
evaluation: CellPLM Norman MSE (Table 5)
evaluation: CellPLM Norman MSE DE (Table 5)
evaluation: CellPLM Norman Pearson (Table 5)
evaluation: CellPLM Norman Pearson DE (Table 5)
evaluation: CellPLM Norman Pearson Delta (Table 5)
evaluation: CellPLM Norman Pearson Delta DE (Table 5)

Suggest a correction