Configuration

scGPT (SCMBench)

Evidence concern: excluded from comparisons

Sheet 'unpaired_inte_overall' has no header for column H. Each H cell is the formula =AVERAGE(Bn:Gn), the mean of the dataset cells with values. Rows with a blank Yao cell (bindSC, scMoMaT, scGPT_zero, Pamona, scFoundation, Geneformer) average five datasets and the other rows six, so column H values are not comparable across all rows.SCMBench Supplementary Data 2 (integration results) · Supplementary Data 2, sheet 'unpaired_inte_overall', row 1 and column H

'FM-scVI' is a different foundation model per dataset ('the best-performing foundation model combined with scVI, which varies across datasets', Fig. 2 legend), so its rows do not describe one fixed configuration.SCMBench Supplementary Data 2 (integration results) · Article Fig. 2 legend; Supplementary Data 2 'FM-scVI' rows

In sheet 'paired_inte_overall' the scGPT row is scored on 10x-Multiome-Pbmc10k only; its Overall cell G20 (=AVERAGE(B20:F20)) is that single value, not a five-dataset mean, so it is not comparable with the other Overall values.SCMBench Supplementary Data 2 (integration results) · Supplementary Data 2, sheet 'paired_inte_overall', cell G20

Configuration of scGPT: scGPT; fine-tuned or zero-shot mode not stated in the sheet; Defaults for R methods and methods without explicit hyperparameters; tuned otherwise.

2 evaluations · 2 results 4 open disagreements between studies involve its model

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.

Evaluations and results

2 evaluations · 2 results. Different protocols are not a single leaderboard.

Protocol: SCMBench paired RNA-ATAC integration accuracy

Sorted by Scmbench integration accuracy (mean of MAP, NMI, ASW and ARI; across-dataset mean over 10x-Multiome-Pbmc10k only) (higher is better). The best value in each column is highlighted. Decimals are rounded for display; each value links to the printed value and its source.

Tested configuration	Protocol and dataset	Scmbench integration accuracy (mean of MAP, NMI, ASW and ARI; across-dataset mean over 10x-Multiome-Pbmc10k only)	Scmbench integration accuracy (mean of MAP, NMI, ASW and ARI)
scGPT (SCMBench)Evaluation	paired-overall (SCMBench)	0.591	Not reported
scGPT (SCMBench)Evaluation	10x-Multiome-Pbmc10k (SCMBench)	Not reported	0.591

All 2 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: scGPT (SCMBench)	Protocol: SCMBench paired RNA-ATAC integration accuracy Dataset subset: 10x-Multiome-Pbmc10k (SCMBench)	0.591 Scmbench integration accuracy (mean of MAP, NMI, ASW and ARI) 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 Sheet 'unpaired_inte_overall' has no header for column H. Each H cell is the formula =AVERAGE(Bn:Gn), the mean of the dataset cells with values. Rows with a blank Yao cell (bindSC, scMoMaT, scGPT_zero, Pamona, scFoundation, Geneformer) average five datasets and the other rows six, so column H values are not comparable across all rows.SCMBench Supplementary Data 2 (integration results) · Supplementary Data 2, sheet 'unpaired_inte_overall', row 1 and column H 'FM-scVI' is a different foundation model per dataset ('the best-performing foundation model combined with scVI, which varies across datasets', Fig. 2 legend), so its rows do not describe one fixed configuration.SCMBench Supplementary Data 2 (integration results) · Article Fig. 2 legend; Supplementary Data 2 'FM-scVI' rows In sheet 'paired_inte_overall' the scGPT row is scored on 10x-Multiome-Pbmc10k only; its Overall cell G20 (=AVERAGE(B20:F20)) is that single value, not a five-dataset mean, so it is not comparable with the other Overall values.SCMBench Supplementary Data 2 (integration results) · Supplementary Data 2, sheet 'paired_inte_overall', cell G20 Methods, coverage and source scGPT on SCMBench paired integration, 10x-Multiome-Pbmc10k lit-cells-networks-20261011-scmbench-protocol-paired-integration-accuracy Aggregation: Not reported SCMBench: benchmarking domain-specific and foundation models for single-cell multi-omics data integration; SCMBench Supplementary Data 2 (integration results) · Supplementary Data 2, sheet 'paired_inte_overall', cell B20 (row 'scGPT', column '10x-Multiome-Pbmc10k')
Configuration: scGPT (SCMBench)	Protocol: SCMBench paired RNA-ATAC integration accuracy Dataset subset: paired-overall (SCMBench)	0.591 Scmbench integration accuracy (mean of MAP, NMI, ASW and ARI; across-dataset mean over 10x-Multiome-Pbmc10k only) 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 Sheet 'unpaired_inte_overall' has no header for column H. Each H cell is the formula =AVERAGE(Bn:Gn), the mean of the dataset cells with values. Rows with a blank Yao cell (bindSC, scMoMaT, scGPT_zero, Pamona, scFoundation, Geneformer) average five datasets and the other rows six, so column H values are not comparable across all rows.SCMBench Supplementary Data 2 (integration results) · Supplementary Data 2, sheet 'unpaired_inte_overall', row 1 and column H 'FM-scVI' is a different foundation model per dataset ('the best-performing foundation model combined with scVI, which varies across datasets', Fig. 2 legend), so its rows do not describe one fixed configuration.SCMBench Supplementary Data 2 (integration results) · Article Fig. 2 legend; Supplementary Data 2 'FM-scVI' rows In sheet 'paired_inte_overall' the scGPT row is scored on 10x-Multiome-Pbmc10k only; its Overall cell G20 (=AVERAGE(B20:F20)) is that single value, not a five-dataset mean, so it is not comparable with the other Overall values.SCMBench Supplementary Data 2 (integration results) · Supplementary Data 2, sheet 'paired_inte_overall', cell G20 Methods, coverage and source scGPT on SCMBench paired integration, paired-overall lit-cells-networks-20261011-scmbench-protocol-paired-integration-accuracy Aggregation: Not reported SCMBench: benchmarking domain-specific and foundation models for single-cell multi-omics data integration; SCMBench Supplementary Data 2 (integration results) · Supplementary Data 2, sheet 'paired_inte_overall', cell G20 (row 'scGPT', column 'Overall')

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

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 2026-10-23-5347695500ed
Property and statement	Original source and location	Review and provenance

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

How to read the evidence

Sources and history

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

2 source records and release history

SCMBench: benchmarking domain-specific and foundation models for single-cell multi-omics data integration · Original source · Published article, Nature Communications 2026 (PMC13342640.1)
SCMBench Supplementary Data 2 (integration results) · Original source · Supplementary Data 2 (MOESM4) of the published article

Download this release (gzip)

Technical metadata and extraction receipts

Stable ID: lit-cells-networks-20261011-scmbench-config-scgpt

areas: single-cell
contexts: research
method types: foundation_model
reported name: scGPT
foundation model eligible: true
hyperparameters: Defaults for R methods and methods without explicit hyperparameters; tuned otherwise (Methods 'Hyperparameter tuning')
source locator: Supplementary Data 2 method rows; Methods 'Benchmarking setup'
missing metadata: version: reason: unreported
provenance note: scGPT as run in SCMBench: scGPT as printed in the paired results (only PBMC-10k scored); fine-tuned or zero-shot mode not stated in the sheet.

Related records

configuration of: scGPT
system: scGPT on SCMBench paired integration, 10x-Multiome-Pbmc10k
system: scGPT on SCMBench paired integration, paired-overall

Suggest a correction