Task

Perturbation response

Perturbation-response prediction asks whether a model can predict molecular measurements under a changed cellular condition.

Sources

altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation

No reviewed evaluations are linked here in this release. See the sources and separately identified configurations below.

0 evaluations · 0 results

Overview

Datasets

No single dataset is fixed by this guide. Select a linked protocol and its versioned data release.

inapplicable

Metrics

PerturBench supports expression-error, correlation, change-from-control and distributional metrics; aggregation and ranking settings are explicit configuration choices.

Allowed inputs

Control-cell measurements, requested condition and permitted covariates; predictions and observed responses aligned to the same features.

Sources

altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation

Evaluation procedure diagram

How it worksConceptual evaluation workflow

Conceptual task guide. Dataset preparation, parameters and scoring must come from a separately identified protocol.

Sources

altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation

Source reviewed · Automated source review, 2026-09-16. All specifications and missing details

Results

All evaluations

0 evaluations · 0 results. Different protocols are not a single leaderboard.

Applied filters: All linked evaluations

No evaluations linked in this release.

Source checking is not independent reproduction. Release 2026-09-29-06401fd5b220.

Methods and evaluation design

Procedure, tasks and evaluated configurations

How it works

Choosing an evaluation

A concrete protocol defines which conditions are observed during fitting and which are held out. Compare predicted and measured responses with matched covariates and a declared aggregation. PerturBench separates average expression, changes from controls and distributional comparisons; these answer different questions. Report the dataset-specific split rather than assuming that a random cell split tests unseen perturbations.

Sources

altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation

Task scope

This is a task guide, not a single versioned benchmark protocol. The connected resources provide examples or concrete procedures. A candidate method or proposed control is not evidence that an evaluation has been completed.

Sources

altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation

Run instructions

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

A task describes a biological question. Choose a linked protocol to obtain concrete split and scoring instructions.

Strengths, limitations and unresolved questions

Strengths and limitations

Strengths supported by sources

Matched controls and held-out conditions allow an evaluation to separate reconstruction from generalisation to a new condition.
Sources
altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation

Limitations and conditions

A good score on average expression need not establish correct changes or cell-to-cell variation. Dose, time and cell context must match.
Sources
altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation

Profile review details

Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol.

Stable record: catalog-task-cell-perturbation

Specifications

Inputs, training, access and other details

Explanatory profile: source reviewed · Automated source review, 2026-09-16. Review applies to the cited claims; unresolved fields are listed below. Numerical results retain their own review status.

Data, procedure and scoring
Property	Description and evidence
Entity type	Task guide; concrete protocol identities remain separate. Sources altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation
Datasets	No single dataset is fixed by this guide. Select a linked protocol and its versioned data release. · Not applicable Sources altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation
Organisms	No shared organism population is defined at this guide level. Record it for each selected dataset. · Not applicable Sources altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation
Assays	No single measurement assay is fixed by this guide; the endpoint and assay belong to the selected protocol. · Not applicable Sources altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation
Splits	No executable split is attached to this task identity. Use the selected protocol’s split manifest. · Not applicable Sources altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation
Allowed inputs	Control-cell measurements, requested condition and permitted covariates; predictions and observed responses aligned to the same features. Sources altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation
Adaptation	No common fitting regime is imposed here. Keep pretrained, frozen, probed, fine-tuned and conventional methods distinct where applicable. · Not applicable Sources altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation
Metrics	PerturBench supports expression-error, correlation, change-from-control and distributional metrics; aggregation and ranking settings are explicit configuration choices. Sources altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation
Baselines	Choose the baselines supplied by the selected protocol and record their access to controls and training conditions. No baseline result is asserted by this guide. Sources altoslabs/perturbench official source · README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation

Evidence

Source checking verifies the cited claim or transcription. It does not establish independent reproduction.

Papers and result coverage

Last literature check: 2026-09-17. Primary-paper discovery and source inspection. Source-checked results are not independently reproduced experiments.

Paper or primary resource	Version	Reference
Predicting transcriptional outcomes of novel multigene perturbations with GEARS	PMC full-text XML snapshot PMC11180609 at retrieval; byte-pinned by SHA-256	Read source DOI: 10.1038/s41587-023-01905-6
Towards Principled Evaluation of Single-Cell Perturbation Prediction Models	bioRxiv 2026.07.23.740433v1	Read source DOI: 10.64898/2026.07.23.740433

Historical gaps recorded on 2026-09-17

A task guide cannot merge perturbation identities, context holdouts, gene subsets or DEG metrics into one leaderboard.
Primary figures/source data must be extracted before adding new numerical comparisons.

Search and extraction details

broad task primary protocol linked

Searches

Open Problems single cell label projection benchmark v1.0.0 paper
CAPRI assessment rounds 46 54 protein docking results
GEARS predicting transcriptional outcomes multigene perturbations 2023
scVI scANVI reference mapping benchmark primary paper
Boltz-2 affinity benchmark paper
ESMFold monomer structure benchmark Science 2023
"ProkBERT" "paper" "2024"
"scPertEval"
ProkBERT family prokaryotic language models microbe paper
Probabilistic harmonization annotation single-cell transcriptomics scANVI Nature Methods 2021
Evolutionary-scale prediction atomic-level protein structure language model ESMFold Science Lin 2023
Towards Principled Evaluation Single-Cell Perturbation Prediction Models Schäfer 2026

Evidence locations

GEARS Results: single/multigene perturbation prediction
GEARS Figure 2; Supplementary Table 2
scPertEval Sections 2–6 and Table 1

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.

17 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-09-29-06401fd5b220
Property and statement	Original source and location	Review and provenance
Diagram caption Conceptual task guide. Dataset preparation, parameters and scoring must come from a separately identified protocol. Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	source checked automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record
Diagram steps Select data and held-out conditions Specify allowed controls and covariates Predict molecular responses Score declared aggregates and distributions Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	source checked automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record
Diagram title Conceptual evaluation workflow Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	source checked automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record
Entity type Task guide; concrete protocol identities remain separate. Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	source checked automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.facts.0.value` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record
Datasets No single dataset is fixed by this guide. Select a linked protocol and its versioned data release. Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	inapplicable automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.facts.1.value` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record
Organisms No shared organism population is defined at this guide level. Record it for each selected dataset. Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	inapplicable automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.facts.2.value` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record
Assays No single measurement assay is fixed by this guide; the endpoint and assay belong to the selected protocol. Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	inapplicable automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.facts.3.value` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record
Splits No executable split is attached to this task identity. Use the selected protocol’s split manifest. Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	inapplicable automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.facts.4.value` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record
Allowed inputs Control-cell measurements, requested condition and permitted covariates; predictions and observed responses aligned to the same features. Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	source checked automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.facts.5.value` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record
Adaptation No common fitting regime is imposed here. Keep pretrained, frozen, probed, fine-tuned and conventional methods distinct where applicable. Individual claims	altoslabs/perturbench official source Original source ↗ README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation Version: c84038bc1ea409aa54f3832cfa6f34f5059adf0c Retrieved: 2026-09-16T10:30:22.733246+00:00	inapplicable automated source review · 2026-09-16 Audit details Reviewed named primary documentation and existing protocol records. Editorial task scope is distinguished from source-reported procedures. No model execution or new numerical result; a reviewed guide is not a complete runnable protocol. Field: `attributes.profile.facts.6.value` Source artifact SHA-256: `4aa29fa1d4333a72013fd2c60545218f0015b85b95b76878cbdb88962ff4e55d` Hash scope: Hash scope not separately documented; inspect source record

All evidence (CSV)All evidence (JSONL)How to read the evidence

Sources and history

View linked audit checks and correction history

Release 2026-09-29-06401fd5b220 · Record review: discovered

6 source records and release history

Geneformer official resource · Original source · Not reported
scFoundation official resource · Original source · Not reported
GEARS official resource · Original source · Not reported
Predicting transcriptional outcomes of novel multigene perturbations with GEARS · Original source · PMC full-text XML snapshot PMC11180609 at retrieval; byte-pinned by SHA-256
Towards Principled Evaluation of Single-Cell Perturbation Prediction Models · Original source · bioRxiv 2026.07.23.740433v1
altoslabs/perturbench official source · Original source · c84038bc1ea409aa54f3832cfa6f34f5059adf0c

Download this release

Technical metadata and extraction receipts

Stable ID: catalog-task-cell-perturbation

areas: cells-tissues
entity level: task
version: Not reported
task: Perturbation response
scope note: Predict expression changes after unseen perturbations.
benchmark research: review date: 2026-09-17; status: broad_task_primary_protocol_linked; primary sources: expansion-p3-gears; expansion-p3-scperteval-paper; inspected locators: GEARS Results: single/multigene perturbation prediction; GEARS Figure 2; Supplementary Table 2; scPertEval Sections 2–6 and Table 1; searched queries: Open Problems single cell label projection benchmark v1.0.0 paper; CAPRI assessment rounds 46 54 protein docking results; GEARS predicting transcriptional outcomes multigene perturbations 2023; scVI scANVI reference mapping benchmark primary paper; Boltz-2 affinity benchmark paper; ESMFold monomer structure benchmark Science 2023; "ProkBERT" "paper" "2024"; "scPertEval"; ProkBERT family prokaryotic language models microbe paper; Probabilistic harmonization annotation single-cell transcriptomics scANVI Nature Methods 2021; Evolutionary-scale prediction atomic-level protein structure language model ESMFold Science Lin 2023; Towards Principled Evaluation Single-Cell Perturbation Prediction Models Schäfer 2026; gaps: A task guide cannot merge perturbation identities, context holdouts, gene subsets or DEG metrics into one leaderboard.; Primary figures/source data must be extracted before adding new numerical comparisons.; claim scope: Primary-paper discovery and source inspection. Source-checked results are not independently reproduced experiments.
historical missing metadata: protocol version: not_yet_extracted
metadata review scope: historical_missing_metadata preserves the original discovery state. Current descriptive evidence and missingness are recorded in profile.facts; numerical-result review is separate.
legacy kinds: benchmark
entity classification: review date: 2026-09-17; rationale: This record identifies the biological prediction question or a suite-specific task, rather than a uniquely fixed evaluated procedure. Preserve its task identity and leave split, model adaptation and scoring details on linked protocols/evaluations.; source ids: src-discovery-altoslabs-perturbench; source locator: README: Downloading and Preparing Datasets, Data Splitting, Evaluator Class and Automated Evaluation; ambiguities: None recorded

Related records

Suggest a correction