Configuration

LA (scGPT)

Baseline implemented by the PerturBench authors.

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

11 evaluations · 20 results. Different protocols are not a single leaderboard.

Sorted by Perturbench cosine logfc (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	Perturbench cosine logfc	Perturbench cosine logfc rank	Perturbench RMSE of mean	Perturbench RMSE rank
LA (scGPT)Evaluation	PerturBench v4 combination prediction of held-out dual perturbations on Norman19, Norman19 (PerturBench v4 combination split)	0.77 ± 4 × 10 − 3 (best in column)	0.0085 ± 1 × 10 − 3 (best in column)	0.044 ± 4 × 10 − 4	0.013 ± 2 × 10 − 3 (best in column)
LA (scGPT)Evaluation	PerturBench CB-COSINE: combination prediction on Norman19, Cosine similarity of log fold change, Norman19 (PerturBench split)	0.77 ± 4 × 10 − 3 (best in column)	Not reported	Not reported	Not reported
LA (scGPT)Evaluation	PerturBench v4 covariate transfer to a held-out cell line on Srivatsan20, Srivatsan20 (PerturBench v4 covariate-transfer split)	0.50 ± 4 × 10 − 3	0.13 ± 7 × 10 − 3	0.017 ± 1 × 10 − 4 (best in column)	0.14 ± 5 × 10 − 3
LA (scGPT)Evaluation	PerturBench CT-COSINE: covariate transfer on Srivatsan20, Cosine similarity of log fold change, Srivatsan20 (PerturBench split)	0.50 ± 4 × 10 − 3	Not reported	Not reported	Not reported
LA (scGPT)Evaluation	PerturBench v4 covariate transfer from mono-culture to co-culture on Frangieh21, Frangieh21 (PerturBench v4 co-culture split)	0.18 ± 6 × 10 − 3	0.27 ± 1 × 10 − 2	0.024 ± 6 × 10 − 5	0.24 ± 1 × 10 − 2
LA (scGPT)Evaluation	PerturBench CB-COSINE-RANK: combination prediction on Norman19, Cosine LogFC rank, Norman19 (PerturBench split)	Not reported	0.0085 ± 1 × 10 − 3 (best in column)	Not reported	Not reported
LA (scGPT)Evaluation	PerturBench CB-RMSE-RANK: combination prediction on Norman19, RMSE mean rank, Norman19 (PerturBench split)	Not reported	Not reported	Not reported	0.013 ± 2 × 10 − 3 (best in column)
LA (scGPT)Evaluation	PerturBench CB-RMSE: combination prediction on Norman19, RMSE of the mean, Norman19 (PerturBench split)	Not reported	Not reported	0.044 ± 4 × 10 − 4	Not reported
LA (scGPT)Evaluation	PerturBench CT-COSINE-RANK: covariate transfer on Srivatsan20, Cosine LogFC rank, Srivatsan20 (PerturBench split)	Not reported	0.13 ± 7 × 10 − 3	Not reported	Not reported
LA (scGPT)Evaluation	PerturBench CT-RMSE-RANK: covariate transfer on Srivatsan20, RMSE mean rank, Srivatsan20 (PerturBench split)	Not reported	Not reported	Not reported	0.14 ± 5 × 10 − 3
LA (scGPT)Evaluation	PerturBench CT-RMSE: covariate transfer on Srivatsan20, RMSE of the mean, Srivatsan20 (PerturBench split)	Not reported	Not reported	0.017 ± 1 × 10 − 4 (best in column)	Not reported

All 20 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: LA (scGPT)	Protocol: PerturBench v4 covariate transfer from mono-culture to co-culture on Frangieh21 Dataset subset: Frangieh21 (PerturBench v4 co-culture split)	0.27 ± 1 × 10 − 2 Perturbench cosine logfc rank fraction · lower Uncertainty: SD 0.01 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 frangieh21 lit-cells-networks-20261011-perturbench-v4-protocol-frangieh21 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 7, row 'LA (scGPT)', column 'Cosine LogFC rank'
Configuration: LA (scGPT)	Protocol: PerturBench v4 covariate transfer from mono-culture to co-culture on Frangieh21 Dataset subset: Frangieh21 (PerturBench v4 co-culture split)	0.18 ± 6 × 10 − 3 Perturbench cosine logfc fraction · higher Uncertainty: SD 0.006 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 frangieh21 lit-cells-networks-20261011-perturbench-v4-protocol-frangieh21 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 7, row 'LA (scGPT)', column 'Cosine logFC'
Configuration: LA (scGPT)	Protocol: PerturBench v4 covariate transfer from mono-culture to co-culture on Frangieh21 Dataset subset: Frangieh21 (PerturBench v4 co-culture split)	0.24 ± 1 × 10 − 2 Perturbench RMSE rank fraction · lower Uncertainty: SD 0.01 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 frangieh21 lit-cells-networks-20261011-perturbench-v4-protocol-frangieh21 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 8, row 'LA (scGPT)', column 'RMSE mean rank'
Configuration: LA (scGPT)	Protocol: PerturBench v4 covariate transfer from mono-culture to co-culture on Frangieh21 Dataset subset: Frangieh21 (PerturBench v4 co-culture split)	0.024 ± 6 × 10 − 5 Perturbench RMSE of mean unit-unreported · lower Uncertainty: SD 0.00006 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 frangieh21 lit-cells-networks-20261011-perturbench-v4-protocol-frangieh21 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 8, row 'LA (scGPT)', column 'RMSE mean'
Configuration: LA (scGPT)	Protocol: PerturBench v4 combination prediction of held-out dual perturbations on Norman19 Dataset subset: Norman19 (PerturBench v4 combination split)	0.0085 ± 1 × 10 − 3 Perturbench cosine logfc rank fraction · lower Uncertainty: SD 0.001 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 norman19 lit-cells-networks-20261011-perturbench-v4-protocol-norman19 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 3, row 'LA (scGPT)', column 'Cosine LogFC rank'
Configuration: LA (scGPT)	Protocol: PerturBench v4 combination prediction of held-out dual perturbations on Norman19 Dataset subset: Norman19 (PerturBench v4 combination split)	0.77 ± 4 × 10 − 3 Perturbench cosine logfc fraction · higher Uncertainty: SD 0.004 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 norman19 lit-cells-networks-20261011-perturbench-v4-protocol-norman19 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 3, row 'LA (scGPT)', column 'Cosine logFC'
Configuration: LA (scGPT)	Protocol: PerturBench v4 combination prediction of held-out dual perturbations on Norman19 Dataset subset: Norman19 (PerturBench v4 combination split)	0.013 ± 2 × 10 − 3 Perturbench RMSE rank fraction · lower Uncertainty: SD 0.002 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 norman19 lit-cells-networks-20261011-perturbench-v4-protocol-norman19 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 6, row 'LA (scGPT)', column 'RMSE mean rank'
Configuration: LA (scGPT)	Protocol: PerturBench v4 combination prediction of held-out dual perturbations on Norman19 Dataset subset: Norman19 (PerturBench v4 combination split)	0.044 ± 4 × 10 − 4 Perturbench RMSE of mean unit-unreported · lower Uncertainty: SD 0.0004 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 norman19 lit-cells-networks-20261011-perturbench-v4-protocol-norman19 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 6, row 'LA (scGPT)', column 'RMSE mean'
Configuration: LA (scGPT)	Protocol: PerturBench v4 covariate transfer to a held-out cell line on Srivatsan20 Dataset subset: Srivatsan20 (PerturBench v4 covariate-transfer split)	0.13 ± 7 × 10 − 3 Perturbench cosine logfc rank fraction · lower Uncertainty: SD 0.007 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 srivatsan20 lit-cells-networks-20261011-perturbench-v4-protocol-srivatsan20 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 2, row 'LA (scGPT)', column 'Cosine LogFC rank'
Configuration: LA (scGPT)	Protocol: PerturBench v4 covariate transfer to a held-out cell line on Srivatsan20 Dataset subset: Srivatsan20 (PerturBench v4 covariate-transfer split)	0.50 ± 4 × 10 − 3 Perturbench cosine logfc fraction · higher Uncertainty: SD 0.004 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 srivatsan20 lit-cells-networks-20261011-perturbench-v4-protocol-srivatsan20 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 2, row 'LA (scGPT)', column 'Cosine LogFC'
Configuration: LA (scGPT)	Protocol: PerturBench v4 covariate transfer to a held-out cell line on Srivatsan20 Dataset subset: Srivatsan20 (PerturBench v4 covariate-transfer split)	0.14 ± 5 × 10 − 3 Perturbench RMSE rank fraction · lower Uncertainty: SD 0.005 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 srivatsan20 lit-cells-networks-20261011-perturbench-v4-protocol-srivatsan20 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 5, row 'LA (scGPT)', column 'RMSE mean rank'
Configuration: LA (scGPT)	Protocol: PerturBench v4 covariate transfer to a held-out cell line on Srivatsan20 Dataset subset: Srivatsan20 (PerturBench v4 covariate-transfer split)	0.017 ± 1 × 10 − 4 Perturbench RMSE of mean unit-unreported · lower Uncertainty: SD 0.0001 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench v4 srivatsan20 lit-cells-networks-20261011-perturbench-v4-protocol-srivatsan20 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 5, row 'LA (scGPT)', column 'RMSE mean'
Configuration: LA (scGPT)	Task: PerturBench CB-COSINE: combination prediction on Norman19, Cosine similarity of log fold change Dataset subset: Norman19 (PerturBench split)	0.77 ± 4 × 10 − 3 Perturbench cosine logfc fraction · higher Uncertainty: SD 0.004 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench CB-COSINE: combination prediction on Norman19, Cosine similarity of log fold change Predict the effect of a pair of gene overexpressions from single perturbations, reported as the mean and one standard deviation over seeds. Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis · Table 3, row(LA (scGPT)), column(Cosine, log fold change (LogFC))
Configuration: LA (scGPT)	Task: PerturBench CB-COSINE-RANK: combination prediction on Norman19, Cosine LogFC rank Dataset subset: Norman19 (PerturBench split)	0.0085 ± 1 × 10 − 3 Perturbench cosine logfc rank fraction · lower Uncertainty: SD 0.001 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench CB-COSINE-RANK: combination prediction on Norman19, Cosine LogFC rank Predict the effect of a pair of gene overexpressions from single perturbations, reported as the mean and one standard deviation over seeds. Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis · Table 3, row(LA (scGPT)), column(Cosine, LogFC rank)
Configuration: LA (scGPT)	Task: PerturBench CB-RMSE-RANK: combination prediction on Norman19, RMSE mean rank Dataset subset: Norman19 (PerturBench split)	0.013 ± 2 × 10 − 3 Perturbench RMSE rank fraction · lower Uncertainty: SD 0.002 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench CB-RMSE-RANK: combination prediction on Norman19, RMSE mean rank Predict the effect of a pair of gene overexpressions from single perturbations, reported as the mean and one standard deviation over seeds. Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis · Table 3, row(LA (scGPT)), column(RMSE, mean rank)
Configuration: LA (scGPT)	Task: PerturBench CB-RMSE: combination prediction on Norman19, RMSE of the mean Dataset subset: Norman19 (PerturBench split)	0.044 ± 4 × 10 − 4 Perturbench RMSE of mean unit-unreported · lower Uncertainty: SD 0.0004 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench CB-RMSE: combination prediction on Norman19, RMSE of the mean Predict the effect of a pair of gene overexpressions from single perturbations, reported as the mean and one standard deviation over seeds. Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis · Table 3, row(LA (scGPT)), column(RMSE, mean)
Configuration: LA (scGPT)	Task: PerturBench CT-COSINE: covariate transfer on Srivatsan20, Cosine similarity of log fold change Dataset subset: Srivatsan20 (PerturBench split)	0.50 ± 4 × 10 − 3 Perturbench cosine logfc fraction · higher Uncertainty: SD 0.004 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench CT-COSINE: covariate transfer on Srivatsan20, Cosine similarity of log fold change Train on some cell types and predict drug effects in a held-out cell type, reported as the mean and one standard deviation over seeds. Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis · Table 2, row(LA (scGPT)), column(Cosine, log fold change (LogFC))
Configuration: LA (scGPT)	Task: PerturBench CT-COSINE-RANK: covariate transfer on Srivatsan20, Cosine LogFC rank Dataset subset: Srivatsan20 (PerturBench split)	0.13 ± 7 × 10 − 3 Perturbench cosine logfc rank fraction · lower Uncertainty: SD 0.007 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench CT-COSINE-RANK: covariate transfer on Srivatsan20, Cosine LogFC rank Train on some cell types and predict drug effects in a held-out cell type, reported as the mean and one standard deviation over seeds. Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis · Table 2, row(LA (scGPT)), column(Cosine, LogFC rank)
Configuration: LA (scGPT)	Task: PerturBench CT-RMSE-RANK: covariate transfer on Srivatsan20, RMSE mean rank Dataset subset: Srivatsan20 (PerturBench split)	0.14 ± 5 × 10 − 3 Perturbench RMSE rank fraction · lower Uncertainty: SD 0.005 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench CT-RMSE-RANK: covariate transfer on Srivatsan20, RMSE mean rank Train on some cell types and predict drug effects in a held-out cell type, reported as the mean and one standard deviation over seeds. Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis · Table 2, row(LA (scGPT)), column(RMSE, mean rank)
Configuration: LA (scGPT)	Task: PerturBench CT-RMSE: covariate transfer on Srivatsan20, RMSE of the mean Dataset subset: Srivatsan20 (PerturBench split)	0.017 ± 1 × 10 − 4 Perturbench RMSE of mean unit-unreported · lower Uncertainty: SD 0.0001 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source LA (scGPT) on PerturBench CT-RMSE: covariate transfer on Srivatsan20, RMSE of the mean Train on some cell types and predict drug effects in a held-out cell type, reported as the mean and one standard deviation over seeds. Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis · Table 2, row(LA (scGPT)), column(RMSE, mean)

Source checking is not independent reproduction. Release 2026-11-02-a7f03c9e55a5.

Use this configuration

How it works, versions and access

Underlying model: scGPT. Results on this page belong to this configuration and its evaluated settings.

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.

1 evidence row matching the loaded filters

Claims, original sources and review scope · Release 2026-11-02-a7f03c9e55a5
Property and statement	Original source and location	Review and provenance
Relationship: uses model catalog-model-scgpt Individual claims	PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis Original source ↗ Section 3.2 Models, paragraph S3.SS2.p1.1 and Appendix E.5 scGPT Embeddings Version: Primary full-text snapshot retrieved 2026-09-17; exact bytes pinned by SHA-256 Retrieved: 2026-09-17T08:06:28.400469+00:00	source checked ["automated-source-review"] · 2026-09-23 Audit details Source review establishes this relationship only. Exact evaluated configurations and original numerical review status remain unchanged. Latent additive predictor consumes scGPT embeddings; results must remain a downstream pipeline. Field: `links:uses_model:catalog-model-scgpt` Claim: model-evaluation-identity-fa7e29e2025566e1e2b7 Source artifact SHA-256: `5c4804565dd9faa17a11853a79e9847dcb6da73715b4c91f62e5b274cc79f186` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact

How to read the evidence

Sources and history

Release 2026-11-02-a7f03c9e55a5 · Record review: source checked

1 source record and release history

PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis · Original source · Primary full-text snapshot retrieved 2026-09-17; exact bytes pinned by SHA-256

Download this release (gzip)

Technical metadata and extraction receipts

Stable ID: perturbench-method-la-scgpt

areas: cells-tissues
source locator: Table 2, row(LA (scGPT))
missing metadata: checkpoint revision: reason: unreported; parameters: reason: unextracted

Related records

Suggest a correction