Evaluator

scPertEval

scPertEval evaluates and calibrates scoring protocols for single-cell perturbation predictions.

Sources

Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy

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

0 evaluations · 0 results

Overview

Datasets

Predicted and observed perturbation responses; the associated study evaluates protocols across public datasets.

Metrics

Protocol scoring is separated from calibration against empirical positive/negative controls using DRF and BDS.

Allowed inputs

Predicted/observed responses plus a protocol specifying representation, metric, transformation and reporting.

Sources

Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy

Evaluation procedure diagram

How it worksEvaluation procedure

Conceptual procedure. Task variants and protocol versions retain their separate scoring conditions.

Sources

Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy

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

Evaluation methodology

scPertEval supplies explicit scoring protocols for perturbation predictions. A metric can compare one perturbation or operate across the complete perturbation panel, depending on its declared scope. Ground-truth references and context are passed to the evaluator, so protocol and preprocessing choices remain part of the result.

Sources

scperteval0 primary benchmark evidence · Pinned src/scperteval/protocols/metrics.py: metric input contract and context

Run instructions

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

Strengths, limitations and unresolved questions

Strengths and limitations

Strengths supported by sources

Calibration tests whether a score distinguishes empirical controls before using it to rank models.
Sources
Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy

Limitations and conditions

Metric computation alone cannot establish whether a predictor saw a held-out perturbation or cell context during training. That evidence must come from the model and dataset protocol.
Sources
scperteval0 primary benchmark evidence · Pinned src/scperteval/protocols/metrics.py: metric input contract and context

Profile review details

Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction.

Stable record: discovery-benchmark-scperteval

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
Datasets	Predicted and observed perturbation responses; the associated study evaluates protocols across public datasets. Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Splits	The evaluator scores supplied outputs; predictor train/test partitions belong to the dataset/run being evaluated. · Not applicable Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Metrics	Protocol scoring is separated from calibration against empirical positive/negative controls using DRF and BDS. Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Baselines	Empirical controls calibrate how well a protocol distinguishes expected response quality. Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Leakage controls	The metric implementation receives ground truth, predictions and an evaluation context; it does not construct model-training partitions or audit training data. Predictor leakage controls belong to the protocol and dataset used to produce the submitted predictions. · Not applicable Sources scperteval0 primary benchmark evidence · Pinned src/scperteval/protocols/metrics.py: metric input contract and context
Uncertainty	Uncertainty across samples, datasets or training runs must be defined by the evaluation study; this evaluator entry does not fix one experiment. · Not applicable Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Entity type	Perturbation scoring-protocol calibration toolkit. Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Organisms	Organism scope belongs to the selected perturbation dataset. · Not applicable Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Assays	Observed single-cell perturbation responses and empirical positive/negative controls. Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Allowed inputs	Predicted/observed responses plus a protocol specifying representation, metric, transformation and reporting. Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Adaptation	The toolkit scores and calibrates evaluation protocols; it does not impose predictor fine-tuning. · Not applicable Sources Virtual-Cell-Research-Community/scPertEval official source · Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy
Implementation	Not extracted or verified for this record.

Applicable tests and references

Applicability is distinct from a completed evaluation.

Experimental replicate agreement · proposed

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
Towards Principled Evaluation of Single-Cell Perturbation Prediction Models	bioRxiv 2026.07.23.740433v1	Read source DOI: 10.64898/2026.07.23.740433
scPertEval documentation	Documentation snapshot at retrieval; byte-pinned by SHA-256	Read source

Historical gaps recorded on 2026-09-17

Supplementary Figure 1 reuses a separate Wei et al. benchmark; its origin must be retained rather than attributed as newly run scPertEval model results.
No universal default metric/normalization or scientific model winner is implied by the evaluator.

Search and extraction details

primary protocol screened

Searches

"PMC13182013"
"PMC12619997"
ProkBERT promoter benchmark
scPertEval benchmark 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

Primary paper abstract, Sections 2–6 and Table 1
Supplementary Figure 1; Supplementary Table 1
Data/code availability

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.

18 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 procedure. Task variants and protocol versions retain their separate scoring conditions. Individual claims	Virtual-Cell-Research-Community/scPertEval official source Original source ↗ Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy Version: 4685f11927e887745737600170da7a655b727553 Retrieved: 2026-09-16T10:30:23.079976+00:00	source checked automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `d8522e585806ec0008a36558e0dd1f6f0b0bb4deffb3f64a7ee4f09cd087e9f4` Hash scope: Hash scope not separately documented; inspect source record
Diagram steps Allowed inputs: Predicted/observed responses plus a protocol specifying representation, metric, transformation and reporting. Splits: The evaluator scores supplied outputs; predictor train/test partitions belong to the dataset/run being evaluated. Metrics: Protocol scoring is separated from calibration against empirical positive/negative controls using DRF and BDS. Individual claims	Virtual-Cell-Research-Community/scPertEval official source Original source ↗ Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy Version: 4685f11927e887745737600170da7a655b727553 Retrieved: 2026-09-16T10:30:23.079976+00:00	source checked automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `d8522e585806ec0008a36558e0dd1f6f0b0bb4deffb3f64a7ee4f09cd087e9f4` Hash scope: Hash scope not separately documented; inspect source record
Diagram title Evaluation procedure Individual claims	Virtual-Cell-Research-Community/scPertEval official source Original source ↗ Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy Version: 4685f11927e887745737600170da7a655b727553 Retrieved: 2026-09-16T10:30:23.079976+00:00	source checked automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `d8522e585806ec0008a36558e0dd1f6f0b0bb4deffb3f64a7ee4f09cd087e9f4` Hash scope: Hash scope not separately documented; inspect source record
Datasets Predicted and observed perturbation responses; the associated study evaluates protocols across public datasets. Individual claims	Virtual-Cell-Research-Community/scPertEval official source Original source ↗ Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy Version: 4685f11927e887745737600170da7a655b727553 Retrieved: 2026-09-16T10:30:23.079976+00:00	source checked automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.facts.0.value` Source artifact SHA-256: `d8522e585806ec0008a36558e0dd1f6f0b0bb4deffb3f64a7ee4f09cd087e9f4` Hash scope: Hash scope not separately documented; inspect source record
Splits The evaluator scores supplied outputs; predictor train/test partitions belong to the dataset/run being evaluated. Individual claims	Virtual-Cell-Research-Community/scPertEval official source Original source ↗ Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy Version: 4685f11927e887745737600170da7a655b727553 Retrieved: 2026-09-16T10:30:23.079976+00:00	inapplicable automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.facts.1.value` Source artifact SHA-256: `d8522e585806ec0008a36558e0dd1f6f0b0bb4deffb3f64a7ee4f09cd087e9f4` Hash scope: Hash scope not separately documented; inspect source record
Adaptation The toolkit scores and calibrates evaluation protocols; it does not impose predictor fine-tuning. Individual claims	Virtual-Cell-Research-Community/scPertEval official source Original source ↗ Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy Version: 4685f11927e887745737600170da7a655b727553 Retrieved: 2026-09-16T10:30:23.079976+00:00	inapplicable automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.facts.10.value` Source artifact SHA-256: `d8522e585806ec0008a36558e0dd1f6f0b0bb4deffb3f64a7ee4f09cd087e9f4` Hash scope: Hash scope not separately documented; inspect source record
Metrics Protocol scoring is separated from calibration against empirical positive/negative controls using DRF and BDS. Individual claims	Virtual-Cell-Research-Community/scPertEval official source Original source ↗ Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy Version: 4685f11927e887745737600170da7a655b727553 Retrieved: 2026-09-16T10:30:23.079976+00:00	source checked automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.facts.2.value` Source artifact SHA-256: `d8522e585806ec0008a36558e0dd1f6f0b0bb4deffb3f64a7ee4f09cd087e9f4` Hash scope: Hash scope not separately documented; inspect source record
Baselines Empirical controls calibrate how well a protocol distinguishes expected response quality. Individual claims	Virtual-Cell-Research-Community/scPertEval official source Original source ↗ Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy Version: 4685f11927e887745737600170da7a655b727553 Retrieved: 2026-09-16T10:30:23.079976+00:00	source checked automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.facts.3.value` Source artifact SHA-256: `d8522e585806ec0008a36558e0dd1f6f0b0bb4deffb3f64a7ee4f09cd087e9f4` Hash scope: Hash scope not separately documented; inspect source record
Leakage controls The metric implementation receives ground truth, predictions and an evaluation context; it does not construct model-training partitions or audit training data. Predictor leakage controls belong to the protocol and dataset used to produce the submitted predictions. Individual claims	scperteval0 primary benchmark evidence Original source ↗ Pinned src/scperteval/protocols/metrics.py: metric input contract and context Version: 4685f11927e887745737600170da7a655b727553:src/scperteval/protocols/metrics.py Retrieved: 2026-09-16T21:11:45.124583+00:00	inapplicable automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.facts.4.value` Source artifact SHA-256: `c4dbfdbc539ba78350ddba68ca4c03edfe89cf3625c6c699884bd3a85e47d929` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Uncertainty Uncertainty across samples, datasets or training runs must be defined by the evaluation study; this evaluator entry does not fix one experiment. Individual claims	Virtual-Cell-Research-Community/scPertEval official source Original source ↗ Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy Version: 4685f11927e887745737600170da7a655b727553 Retrieved: 2026-09-16T10:30:23.079976+00:00	inapplicable automated source review · 2026-09-16 Audit details Primary paper and/or task implementation reviewed for the explicitly cited methodology claims. Scope-limited absence is recorded only after the documented source search; no model runs or independent reproduction. Field: `attributes.profile.facts.5.value` Source artifact SHA-256: `d8522e585806ec0008a36558e0dd1f6f0b0bb4deffb3f64a7ee4f09cd087e9f4` 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

4 source records and release history

Virtual-Cell-Research-Community/scPertEval official source · Original source · 4685f11927e887745737600170da7a655b727553
scPertEval documentation · Original source · Documentation snapshot at retrieval; byte-pinned by SHA-256
Towards Principled Evaluation of Single-Cell Perturbation Prediction Models · Original source · bioRxiv 2026.07.23.740433v1
scperteval0 primary benchmark evidence · Original source · 4685f11927e887745737600170da7a655b727553:src/scperteval/protocols/metrics.py

Download this release

Technical metadata and extraction receipts

Stable ID: discovery-benchmark-scperteval

areas: single-cell
entity level: evaluator
scope note: Specialist molecular or omics evaluation; protocol details require review before numerical comparison.
task: Perturbation prediction metric calibration
version: Not reported
benchmark research: review date: 2026-09-17; status: primary_protocol_screened; primary sources: expansion-p3-scperteval-paper; expansion-p3-scperteval-docs; inspected locators: Primary paper abstract, Sections 2–6 and Table 1; Supplementary Figure 1; Supplementary Table 1; Data/code availability; searched queries: "PMC13182013"; "PMC12619997"; ProkBERT promoter benchmark; scPertEval benchmark 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: Supplementary Figure 1 reuses a separate Wei et al. benchmark; its origin must be retained rather than attributed as newly run scPertEval model results.; No universal default metric/normalization or scientific model winner is implied by the evaluator.; claim scope: Primary-paper discovery and source inspection. Source-checked results are not independently reproduced experiments.
historical missing metadata: dataset release: unextracted; metric implementation: unextracted; split manifest: unextracted; version: unextracted
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: The cited profile describes scoring or assessment software/procedures applied to predictions, rather than the biological target or the underlying dataset.; source ids: src-discovery-virtual-cell-research-community-scperteval; source locator: Pinned README: toolkit purpose; scoring/calibration/DE actions; protocol taxonomy; ambiguities: None recorded

Related records

applicable to: Experimental replicate agreement

Suggest a correction