Protocol

PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3

PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3; mean and one standard deviation over 5 seeds of the best HPO configuration.

8 evaluations · 55 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.

8 recorded evaluations, 55 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

8 evaluations · 55 results. Different protocols are not a single leaderboard.

Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 · Dataset subset: OP3 (PerturBench v4 split)

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	Perturbench cosine logfc	Perturbench cosine logfc rank	Perturbench energy mmd (gene expression space)	Perturbench energy mmd (PCA space: top 256 principal components fitted on the ground-truth test split)	Perturbench RMSE of mean	Perturbench RMSE rank	Recall (DEG recall: fraction of the top 20 ground-truth DEGs (scanpy t-scores against controls) recalled from predicted cells)
SAMS-VAE* (S)Evaluation	0.43 ± 2 × 10 − 3 (best in column)	0.079 ± 7 × 10 − 3	5.6 ± 9 × 10 − 3	0.74 ± 3 × 10 − 3	0.031 ± 1 × 10 − 4 (best in column)	0.051 ± 4 × 10 − 3 (best in column)	0.000 ± 0
SAMS-VAE*Evaluation	0.41 ± 1 × 10 − 2	0.11 ± 2 × 10 − 2	5.6 ± 4 × 10 − 2	0.90 ± 3 × 10 − 2	0.034 ± 5 × 10 − 4	0.11 ± 2 × 10 − 2	0.000 ± 0
CPA*Evaluation	0.39 ± 6 × 10 − 3	0.091 ± 2 × 10 − 3	5.3 ± 3 × 10 − 2	0.71 ± 1 × 10 − 2 (best in column)	0.032 ± 2 × 10 − 4	0.062 ± 4 × 10 − 3	0.002 ± 2 × 10 − 4 (best in column)
LAEvaluation	0.39 ± 2 × 10 − 2	0.12 ± 3 × 10 − 2	11.0 ± 6 × 10 − 2	4.1 ± 5 × 10 − 2	0.036 ± 2 × 10 − 3	0.14 ± 3 × 10 − 2	0.000 ± 6 × 10 − 5
CPA* (noAdv)Evaluation	0.32 ± 6 × 10 − 3	0.076 ± 3 × 10 − 3	5.7 ± 1 × 10 − 2	0.86 ± 1 × 10 − 2	0.034 ± 4 × 10 − 4	0.051 ± 4 × 10 − 3 (best in column)	0.000 ± 0
DecoderEvaluation	0.32 ± 6 × 10 − 3	0.070 ± 5 × 10 − 3	11.0 ± 5 × 10 − 3	4.4 ± 1 × 10 − 2	0.035 ± 2 × 10 − 4	0.068 ± 6 × 10 − 3	Not reported
Biolord*Evaluation	0.28 ± 3 × 10 − 3	0.051 ± 5 × 10 − 3 (best in column)	4.6 ± 2 × 10 − 2	0.92 ± 5 × 10 − 3	0.038 ± 8 × 10 − 5	0.051 ± 6 × 10 − 3 (best in column)	0.000 ± 0
LinearEvaluation	0.011 ± 9 × 10 − 5	0.14 ± 4 × 10 − 4	4.0 ± 7 × 10 − 3 (best in column)	2.6 ± 4 × 10 − 3	0.070 ± 4 × 10 − 5	0.24 ± 9 × 10 − 4	0.000 ± 0

All 55 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: Biolord*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.051 ± 5 × 10 − 3 Perturbench cosine logfc rank fraction · lower Uncertainty: SD 0.005 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Biolord ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'Biolord ∗', column 'Cosine LogFC rank'
Configuration: Biolord*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.28 ± 3 × 10 − 3 Perturbench cosine logfc fraction · higher Uncertainty: SD 0.003 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Biolord ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'Biolord ∗', column 'Cosine LogFC'
Configuration: Biolord*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.000 ± 0 Recall (DEG recall: fraction of the top 20 ground-truth DEGs (scanpy t-scores against controls) recalled from predicted cells) fraction · higher Uncertainty: SD 0 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Biolord ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'Biolord ∗', column 'DEG recall'
Configuration: Biolord*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.92 ± 5 × 10 − 3 Perturbench energy mmd (PCA space: top 256 principal components fitted on the ground-truth test split) unit-unreported · lower Uncertainty: SD 0.005 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Biolord ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'Biolord ∗', column 'MMD PCA'
Configuration: Biolord*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	4.6 ± 2 × 10 − 2 Perturbench energy mmd (gene expression space) unit-unreported · lower Uncertainty: SD 0.02 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Biolord ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'Biolord ∗', column 'MMD'
Configuration: Biolord*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.051 ± 6 × 10 − 3 Perturbench RMSE rank fraction · lower Uncertainty: SD 0.006 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Biolord ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'Biolord ∗', column 'RMSE mean rank'
Configuration: Biolord*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.038 ± 8 × 10 − 5 Perturbench RMSE of mean unit-unreported · lower Uncertainty: SD 0.00008 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source Biolord ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'Biolord ∗', column 'RMSE mean'
Configuration: CPA*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.091 ± 2 × 10 − 3 Perturbench cosine logfc rank fraction · lower Uncertainty: SD 0.002 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source CPA ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'CPA ∗', column 'Cosine LogFC rank'
Configuration: CPA*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.39 ± 6 × 10 − 3 Perturbench cosine logfc fraction · higher Uncertainty: SD 0.006 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source CPA ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'CPA ∗', column 'Cosine LogFC'
Configuration: CPA*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.002 ± 2 × 10 − 4 Recall (DEG recall: fraction of the top 20 ground-truth DEGs (scanpy t-scores against controls) recalled from predicted cells) fraction · higher Uncertainty: SD 0.0002 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source CPA ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'CPA ∗', column 'DEG recall'
Configuration: CPA*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.71 ± 1 × 10 − 2 Perturbench energy mmd (PCA space: top 256 principal components fitted on the ground-truth test split) unit-unreported · lower Uncertainty: SD 0.01 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source CPA ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'CPA ∗', column 'MMD PCA'
Configuration: CPA*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	5.3 ± 3 × 10 − 2 Perturbench energy mmd (gene expression space) unit-unreported · lower Uncertainty: SD 0.03 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source CPA ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'CPA ∗', column 'MMD'
Configuration: CPA* (noAdv)	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.076 ± 3 × 10 − 3 Perturbench cosine logfc rank fraction · lower Uncertainty: SD 0.003 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source CPA ∗ (noAdv) on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'CPA ∗ (noAdv)', column 'Cosine LogFC rank'
Configuration: CPA* (noAdv)	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.32 ± 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 CPA ∗ (noAdv) on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'CPA ∗ (noAdv)', column 'Cosine LogFC'
Configuration: CPA* (noAdv)	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.000 ± 0 Recall (DEG recall: fraction of the top 20 ground-truth DEGs (scanpy t-scores against controls) recalled from predicted cells) fraction · higher Uncertainty: SD 0 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source CPA ∗ (noAdv) on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'CPA ∗ (noAdv)', column 'DEG recall'
Configuration: CPA* (noAdv)	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.86 ± 1 × 10 − 2 Perturbench energy mmd (PCA space: top 256 principal components fitted on the ground-truth test split) unit-unreported · lower Uncertainty: SD 0.01 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source CPA ∗ (noAdv) on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'CPA ∗ (noAdv)', column 'MMD PCA'
Configuration: CPA* (noAdv)	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	5.7 ± 1 × 10 − 2 Perturbench energy mmd (gene expression space) unit-unreported · lower Uncertainty: SD 0.01 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source CPA ∗ (noAdv) on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'CPA ∗ (noAdv)', column 'MMD'
Configuration: CPA* (noAdv)	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.051 ± 4 × 10 − 3 Perturbench RMSE rank fraction · lower Uncertainty: SD 0.004 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source CPA ∗ (noAdv) on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'CPA ∗ (noAdv)', column 'RMSE mean rank'
Configuration: CPA* (noAdv)	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.034 ± 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 CPA ∗ (noAdv) on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'CPA ∗ (noAdv)', column 'RMSE mean'
Configuration: CPA*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.062 ± 4 × 10 − 3 Perturbench RMSE rank fraction · lower Uncertainty: SD 0.004 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source CPA ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'CPA ∗', column 'RMSE mean rank'
Configuration: CPA*	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.032 ± 2 × 10 − 4 Perturbench RMSE of mean unit-unreported · lower Uncertainty: SD 0.0002 Coverage: Not reported scored / Not reported eligible	Independent external evaluation · Source checked Methods, coverage and source CPA ∗ on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'CPA ∗', column 'RMSE mean'
Configuration: Decoder	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.070 ± 5 × 10 − 3 Perturbench cosine logfc rank fraction · lower Uncertainty: SD 0.005 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source Decoder on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'Decoder', column 'Cosine LogFC rank'
Configuration: Decoder	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	0.32 ± 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 Decoder on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'Decoder', column 'Cosine LogFC'
Configuration: Decoder	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	4.4 ± 1 × 10 − 2 Perturbench energy mmd (PCA space: top 256 principal components fitted on the ground-truth test split) unit-unreported · lower Uncertainty: SD 0.01 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source Decoder on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 9, row 'Decoder', column 'MMD PCA'
Configuration: Decoder	Protocol: PerturBench v4 covariate transfer of small-molecule effects across PBMC cell types on OP3 Dataset subset: OP3 (PerturBench v4 split)	11.0 ± 5 × 10 − 3 Perturbench energy mmd (gene expression space) unit-unreported · lower Uncertainty: SD 0.005 Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source Decoder on PerturBench v4 op3 lit-cells-networks-20261011-perturbench-v4-protocol-op3 Aggregation: Not reported PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Table 10, row 'Decoder', column 'MMD'

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.

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: 5
External evaluations: 3

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

Null control

Proposed control: requires review

No-change prediction under matched control conditions

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

Training-only mean-effect or linear prediction

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

1 source record and release history

PerturBench: Benchmarking Machine Learning Models for Cellular Perturbation Analysis (arXiv v4) · Original source · arXiv 2408.10609v4, posted 2025-10-24 (NeurIPS 2025 Datasets and Benchmarks Track version)

Download this release (gzip)

Technical metadata and extraction receipts

Stable ID: lit-cells-networks-20261011-perturbench-v4-protocol-op3

areas: single-cell
contexts: research
protocol: covariate transfer of small-molecule effects across PBMC cell types on OP3. Hyperparameters chosen by Optuna HPO on RMSE + 0.1 x RMSE rank, then 5 training seeds; metrics computed on the held-out test perturbations.
version: PerturBench arXiv v4
metric: perturbench-cosine-logfc
metric direction: higher
unit: fraction
random seeds: 5 seeds (Table 2 caption; Section 4.3)
limitations: All models were trained and tuned by the PerturBench authors, who also developed the Linear, Latent Additive and Decoder baselines.; Metric columns differ between the main and additional tables; compare only within one metric.
source locator: Tables 9, 10

Related records

part of: PerturBench
uses data: OP3 (PerturBench v4 split)
assessment: Biolord ∗ on PerturBench v4 op3
assessment: CPA ∗ on PerturBench v4 op3
assessment: CPA ∗ (noAdv) on PerturBench v4 op3
assessment: Decoder on PerturBench v4 op3
assessment: LA on PerturBench v4 op3
assessment: Linear on PerturBench v4 op3
assessment: SAMS-VAE ∗ on PerturBench v4 op3
assessment: SAMS-VAE ∗ (S) on PerturBench v4 op3

Suggest a correction