Protocol

SCMBench unpaired RNA-ATAC integration accuracy

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

Integrate unpaired scRNA-seq and scATAC-seq; score the joint embedding by the mean of mean average precision, normalised mutual information, average silhouette width and ARI of cell type labels (integration accuracy overall score).

113 evaluations · 113 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.

113 recorded evaluations, 113 metric rows. A comparison chart has not yet been validated for these results. The table retains the individual findings and their sources.

View coverage and remaining gaps across all benchmarks

Results

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

All evaluations

113 evaluations · 113 results. Different protocols are not a single leaderboard.

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: bindSC (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.671 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 bindSC on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C6 (row 'bindSC', column 'Muto-2021-batch-1')
Configuration: Cobolt (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.315 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 Cobolt on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C14 (row 'Cobolt', column 'Muto-2021-batch-1')
Configuration: FM-scVI (SCMBench adapter)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.576 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	Author-reported 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 FM-scVI on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C18 (row 'FM-scVI', column 'Muto-2021-batch-1')
Configuration: Geneformer (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.356 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 Geneformer on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C15 (row 'Geneformer', column 'Muto-2021-batch-1')
Configuration: GLUE (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.707 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 GLUE on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C4 (row 'GLUE', column 'Muto-2021-batch-1')
Configuration: Harmony (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.603 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 Harmony on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C17 (row 'Harmony', column 'Muto-2021-batch-1')
Configuration: iNMF (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.244 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 iNMF on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C16 (row 'iNMF', column 'Muto-2021-batch-1')
Configuration: LIGER (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.705 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 LIGER on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C5 (row 'LIGER', column 'Muto-2021-batch-1')
Configuration: Pamona (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.51 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 Pamona on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C11 (row 'Pamona', column 'Muto-2021-batch-1')
Configuration: PCA (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.409 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 PCA on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C12 (row 'PCA', column 'Muto-2021-batch-1')
Configuration: scFoundation (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.418 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 scFoundation on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C13 (row 'scFoundation', column 'Muto-2021-batch-1')
Configuration: scGPT_zero (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.534 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_zero on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C10 (row 'scGPT_zero', column 'Muto-2021-batch-1')
Configuration: scJoint (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.812 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 scJoint on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C2 (row 'scJoint', column 'Muto-2021-batch-1')
Configuration: scMoMaT (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.544 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 scMoMaT on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C8 (row 'scMoMaT', column 'Muto-2021-batch-1')
Configuration: scVI (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.579 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 scVI on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C9 (row 'scVI', column 'Muto-2021-batch-1')
Configuration: Seurat4 (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.724 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 Seurat4 on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C3 (row 'Seurat4', column 'Muto-2021-batch-1')
Configuration: UCE (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-1 (SCMBench)	0.612 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 UCE on SCMBench unpaired integration, Muto-2021-batch-1 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell C7 (row 'UCE', column 'Muto-2021-batch-1')
Configuration: bindSC (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-2 (SCMBench)	0.716 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 bindSC on SCMBench unpaired integration, Muto-2021-batch-2 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell D6 (row 'bindSC', column 'Muto-2021-batch-2')
Configuration: Cobolt (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-2 (SCMBench)	0.373 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 Cobolt on SCMBench unpaired integration, Muto-2021-batch-2 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell D14 (row 'Cobolt', column 'Muto-2021-batch-2')
Configuration: FM-scVI (SCMBench adapter)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-2 (SCMBench)	0.599 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	Author-reported 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 FM-scVI on SCMBench unpaired integration, Muto-2021-batch-2 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell D18 (row 'FM-scVI', column 'Muto-2021-batch-2')
Configuration: Geneformer (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-2 (SCMBench)	0.382 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 Geneformer on SCMBench unpaired integration, Muto-2021-batch-2 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell D15 (row 'Geneformer', column 'Muto-2021-batch-2')
Configuration: GLUE (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-2 (SCMBench)	0.713 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 GLUE on SCMBench unpaired integration, Muto-2021-batch-2 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell D4 (row 'GLUE', column 'Muto-2021-batch-2')
Configuration: Harmony (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-2 (SCMBench)	0.631 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 Harmony on SCMBench unpaired integration, Muto-2021-batch-2 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell D17 (row 'Harmony', column 'Muto-2021-batch-2')
Configuration: iNMF (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-2 (SCMBench)	0.32 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 iNMF on SCMBench unpaired integration, Muto-2021-batch-2 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell D16 (row 'iNMF', column 'Muto-2021-batch-2')
Configuration: LIGER (SCMBench)	Protocol: SCMBench unpaired RNA-ATAC integration accuracy Dataset subset: Muto-2021-batch-2 (SCMBench)	0.749 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 LIGER on SCMBench unpaired integration, Muto-2021-batch-2 lit-cells-networks-20261011-scmbench-protocol-unpaired-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 'unpaired_inte_overall', cell D5 (row 'LIGER', column 'Muto-2021-batch-2')

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

Methods and evaluation design

Procedure, tasks and evaluated configurations

Recorded evaluations

Each evaluation records what was tested and under which conditions.

Explore all linked results

Baseline coverage

Reference methods help show what a model adds beyond simple controls. We track a null control and a conventional method for each protocol.

0 of 2 active baseline roles have published Rewire measurements in this release. Measurements on a selected protocol do not establish coverage of an entire suite.

No execution recipe linked to this protocol. Recipe availability does not establish a completed evaluation.

Author-reported evaluations: 7
External evaluations: 106

Literature evidence is not a Rewire measurement. Executed but unpublished runs and private review status are not included.

Null control

Proposed control: requires review

Unintegrated representation using protocol preprocessing

Protocol-specific applicability, permitted inputs, access, split, evaluator and execution requirements need review before implementation or execution.

This is a suggested selection rule, not a validated method or a measured score.

Conventional reference

Proposed control: requires review

Protocol-specific conventional integration method

Protocol-specific applicability, permitted inputs, access, split, evaluator and execution requirements need review before implementation or execution.

This is a suggested selection rule, not a validated method or a measured score.

Protocol coverage CSV (gzip) · Model evaluation matrix (gzip) · Source table (gzip) · Release and checksums (gzip)

Coverage is derived from release 2026-10-23-5347695500ed. Source citations describe the original records; they do not validate an unreviewed baseline proposal. No results have been generated by this audit.

Run instructions

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

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-protocol-unpaired-integration-accuracy

areas: single-cell
contexts: research
protocol: Unpaired multi-omics integration; per-dataset overall score is the mean of MAP, NMI, ASW and ARI ('The overall score represents the mean of these four integration accuracy metrics', Results).
version: Nature Communications 2026
metric: scmbench-integration-accuracy
metric direction: higher
unit: unitless
limitations: Single run per method and dataset; no spread printed.; Foundation models receive ATAC as gene activity scores (MAESTRO).
source locator: Results 'Benchmarking data integration on both paired and unpaired datasets'; Fig. 2e-f; Supplementary Data 2

Related records

part of: SCMBench
assessment: bindSC on SCMBench unpaired integration, Muto-2021-batch-1
assessment: Cobolt on SCMBench unpaired integration, Muto-2021-batch-1
assessment: FM-scVI on SCMBench unpaired integration, Muto-2021-batch-1
assessment: Geneformer on SCMBench unpaired integration, Muto-2021-batch-1
assessment: GLUE on SCMBench unpaired integration, Muto-2021-batch-1
assessment: Harmony on SCMBench unpaired integration, Muto-2021-batch-1
assessment: iNMF on SCMBench unpaired integration, Muto-2021-batch-1
assessment: LIGER on SCMBench unpaired integration, Muto-2021-batch-1
assessment: Pamona on SCMBench unpaired integration, Muto-2021-batch-1
assessment: PCA on SCMBench unpaired integration, Muto-2021-batch-1
assessment: scFoundation on SCMBench unpaired integration, Muto-2021-batch-1
assessment: scGPT_zero on SCMBench unpaired integration, Muto-2021-batch-1
assessment: scJoint on SCMBench unpaired integration, Muto-2021-batch-1
assessment: scMoMaT on SCMBench unpaired integration, Muto-2021-batch-1
assessment: scVI on SCMBench unpaired integration, Muto-2021-batch-1
assessment: Seurat4 on SCMBench unpaired integration, Muto-2021-batch-1
assessment: UCE on SCMBench unpaired integration, Muto-2021-batch-1
assessment: bindSC on SCMBench unpaired integration, Muto-2021-batch-2
assessment: Cobolt on SCMBench unpaired integration, Muto-2021-batch-2
assessment: FM-scVI on SCMBench unpaired integration, Muto-2021-batch-2
assessment: Geneformer on SCMBench unpaired integration, Muto-2021-batch-2
assessment: GLUE on SCMBench unpaired integration, Muto-2021-batch-2
assessment: Harmony on SCMBench unpaired integration, Muto-2021-batch-2
assessment: iNMF on SCMBench unpaired integration, Muto-2021-batch-2
assessment: LIGER on SCMBench unpaired integration, Muto-2021-batch-2
assessment: Pamona on SCMBench unpaired integration, Muto-2021-batch-2
assessment: PCA on SCMBench unpaired integration, Muto-2021-batch-2
assessment: scFoundation on SCMBench unpaired integration, Muto-2021-batch-2
assessment: scGPT_zero on SCMBench unpaired integration, Muto-2021-batch-2
assessment: scJoint on SCMBench unpaired integration, Muto-2021-batch-2
assessment: scMoMaT on SCMBench unpaired integration, Muto-2021-batch-2
assessment: scVI on SCMBench unpaired integration, Muto-2021-batch-2
assessment: Seurat4 on SCMBench unpaired integration, Muto-2021-batch-2
assessment: UCE on SCMBench unpaired integration, Muto-2021-batch-2
assessment: bindSC on SCMBench unpaired integration, Muto-2021-batch-3
assessment: Cobolt on SCMBench unpaired integration, Muto-2021-batch-3
assessment: FM-scVI on SCMBench unpaired integration, Muto-2021-batch-3
assessment: Geneformer on SCMBench unpaired integration, Muto-2021-batch-3
assessment: GLUE on SCMBench unpaired integration, Muto-2021-batch-3
assessment: Harmony on SCMBench unpaired integration, Muto-2021-batch-3
assessment: iNMF on SCMBench unpaired integration, Muto-2021-batch-3
assessment: LIGER on SCMBench unpaired integration, Muto-2021-batch-3
assessment: Pamona on SCMBench unpaired integration, Muto-2021-batch-3
assessment: PCA on SCMBench unpaired integration, Muto-2021-batch-3
assessment: scFoundation on SCMBench unpaired integration, Muto-2021-batch-3
assessment: scGPT_zero on SCMBench unpaired integration, Muto-2021-batch-3
assessment: scJoint on SCMBench unpaired integration, Muto-2021-batch-3
assessment: scMoMaT on SCMBench unpaired integration, Muto-2021-batch-3
assessment: scVI on SCMBench unpaired integration, Muto-2021-batch-3
assessment: Seurat4 on SCMBench unpaired integration, Muto-2021-batch-3
assessment: UCE on SCMBench unpaired integration, Muto-2021-batch-3
assessment: bindSC on SCMBench unpaired integration, Muto-2021-batch-4
assessment: Cobolt on SCMBench unpaired integration, Muto-2021-batch-4
assessment: FM-scVI on SCMBench unpaired integration, Muto-2021-batch-4
assessment: Geneformer on SCMBench unpaired integration, Muto-2021-batch-4
assessment: GLUE on SCMBench unpaired integration, Muto-2021-batch-4
assessment: Harmony on SCMBench unpaired integration, Muto-2021-batch-4
assessment: iNMF on SCMBench unpaired integration, Muto-2021-batch-4
assessment: LIGER on SCMBench unpaired integration, Muto-2021-batch-4
assessment: Pamona on SCMBench unpaired integration, Muto-2021-batch-4
assessment: PCA on SCMBench unpaired integration, Muto-2021-batch-4
assessment: scFoundation on SCMBench unpaired integration, Muto-2021-batch-4
assessment: scGPT_zero on SCMBench unpaired integration, Muto-2021-batch-4
assessment: scJoint on SCMBench unpaired integration, Muto-2021-batch-4
assessment: scMoMaT on SCMBench unpaired integration, Muto-2021-batch-4
assessment: scVI on SCMBench unpaired integration, Muto-2021-batch-4
assessment: Seurat4 on SCMBench unpaired integration, Muto-2021-batch-4
assessment: UCE on SCMBench unpaired integration, Muto-2021-batch-4
assessment: bindSC on SCMBench unpaired integration, Muto-2021-batch-5
assessment: Cobolt on SCMBench unpaired integration, Muto-2021-batch-5
assessment: FM-scVI on SCMBench unpaired integration, Muto-2021-batch-5
assessment: Geneformer on SCMBench unpaired integration, Muto-2021-batch-5
assessment: GLUE on SCMBench unpaired integration, Muto-2021-batch-5
assessment: Harmony on SCMBench unpaired integration, Muto-2021-batch-5
assessment: iNMF on SCMBench unpaired integration, Muto-2021-batch-5
assessment: LIGER on SCMBench unpaired integration, Muto-2021-batch-5
assessment: Pamona on SCMBench unpaired integration, Muto-2021-batch-5
assessment: PCA on SCMBench unpaired integration, Muto-2021-batch-5
assessment: scFoundation on SCMBench unpaired integration, Muto-2021-batch-5
assessment: scGPT_zero on SCMBench unpaired integration, Muto-2021-batch-5
assessment: scJoint on SCMBench unpaired integration, Muto-2021-batch-5
assessment: scMoMaT on SCMBench unpaired integration, Muto-2021-batch-5
assessment: scVI on SCMBench unpaired integration, Muto-2021-batch-5
assessment: Seurat4 on SCMBench unpaired integration, Muto-2021-batch-5
assessment: UCE on SCMBench unpaired integration, Muto-2021-batch-5
assessment: bindSC on SCMBench unpaired integration, unpaired-overall
assessment: Cobolt on SCMBench unpaired integration, unpaired-overall
assessment: FM-scVI on SCMBench unpaired integration, unpaired-overall
assessment: Geneformer on SCMBench unpaired integration, unpaired-overall
assessment: GLUE on SCMBench unpaired integration, unpaired-overall
assessment: Harmony on SCMBench unpaired integration, unpaired-overall
assessment: iNMF on SCMBench unpaired integration, unpaired-overall
assessment: LIGER on SCMBench unpaired integration, unpaired-overall
assessment: Pamona on SCMBench unpaired integration, unpaired-overall
assessment: PCA on SCMBench unpaired integration, unpaired-overall
assessment: scFoundation on SCMBench unpaired integration, unpaired-overall
assessment: scGPT_zero on SCMBench unpaired integration, unpaired-overall
assessment: scJoint on SCMBench unpaired integration, unpaired-overall
assessment: scMoMaT on SCMBench unpaired integration, unpaired-overall
assessment: scVI on SCMBench unpaired integration, unpaired-overall
assessment: Seurat4 on SCMBench unpaired integration, unpaired-overall
assessment: UCE on SCMBench unpaired integration, unpaired-overall
assessment: Cobolt on SCMBench unpaired integration, Yao
assessment: FM-scVI on SCMBench unpaired integration, Yao
assessment: GLUE on SCMBench unpaired integration, Yao
assessment: Harmony on SCMBench unpaired integration, Yao
assessment: iNMF on SCMBench unpaired integration, Yao
assessment: LIGER on SCMBench unpaired integration, Yao
assessment: PCA on SCMBench unpaired integration, Yao
assessment: scJoint on SCMBench unpaired integration, Yao
assessment: scVI on SCMBench unpaired integration, Yao
assessment: Seurat4 on SCMBench unpaired integration, Yao
assessment: UCE on SCMBench unpaired integration, Yao

Suggest a correction