Validating virtual cells: benchmarks, baselines and critiques虚拟细胞的验证：基准、基线与批评
Review pending待审阅 Version 0.1.0 · status: draft · last expert review: not yet reviewed版本 0.1.0 · 状态：初稿 · 最近专家审阅：尚未审阅
Research question研究问题
- How are virtual cell models validated?虚拟细胞模型如何验证？
- Do deep learning models beat linear baselines for perturbation prediction?深度学习模型在扰动预测上能超过线性基线吗？
Current understanding当前认识
Validation currently relies on benchmarks and challenges. Several independent studies report that deep perturbation models do not yet reliably beat simple linear baselines; clinical validation of virtual cells in neuro-oncology has not been reported in the library.目前的验证主要依靠基准与挑战赛。多项独立研究报告深度扰动模型尚不能稳定超越简单线性基线；文献库中尚无神经肿瘤领域虚拟细胞的临床验证报告。
Ahlmann-Eltze et al. (Nature Methods, 2025) report that deep-learning gene perturbation effect prediction does not yet outperform simple linear baselines; a benchmark of foundation cell models and zero-shot evaluations point the same way.Ahlmann-Eltze 等（Nature Methods，2025）报告深度学习基因扰动效应预测尚未超越简单线性基线；基础细胞模型基准与零样本评估指向同一方向。 VC0048 VC0049 VC0050
Evaluation design matters: Systema shows systematic variation can lead to overestimated performance, and preprints propose evaluation principles and report a persisting baseline gap.评估设计至关重要：Systema 显示系统性变异会导致性能高估；多篇预印本提出了评估原则并报告基线差距依然存在。 VC0052 VC0053 · preprint · 预印本 VC0054 · preprint · 预印本
The Virtual Cell Challenge was launched as a “Turing test” for virtual cells; its 2026 edition benchmarks zero-shot generalisation across cellular contexts.虚拟细胞挑战赛作为虚拟细胞的“图灵测试”发起，2026 年赛事评测跨细胞情境的零样本泛化。 VC0002 VC0058
In neuro-oncology imaging, a generative MRI growth-forecasting study in pediatric DMG drew a published critique on domain shift and statistical nesting — a reminder that validation needs independent scrutiny.在神经肿瘤影像领域，一项儿童 DMG 生成式 MRI 生长预测研究受到关于领域偏移与统计嵌套的公开批评——提醒验证需要独立审视。 VC0077 VC0078
Evidence map证据图谱
Primary studies & datasets原始研究与数据集
- VC0048 Deep-learning-based gene perturbation effect prediction does not yet outperform simple linear baselines. — Ahlmann-Eltze C, Huber W, Anders S., Nature methods, 2025
- VC0049 Benchmarking foundation cell models for post-perturbation RNA-seq prediction. — Csendes G, Sanz G, Szalay KZ, Szalai B., BMC genomics, 2025
- VC0050 Zero-shot evaluation reveals limitations of single-cell foundation models. — Kedzierska KZ, Crawford L, Amini AP, Lu AX., Genome biology, 2025
- VC0052 Systema: a framework for evaluating genetic perturbation response prediction beyond systematic variation. — Viñas Torné R, Wiatrak M, Piran Z, Fan S, Jiang L, Teichmann SA, Nitzan M, Brbić M., Nature biotechnology, 2025
- VC0056 Systematic evaluation of single-cell foundation model interpretability: attention-derived edge scores add no incremental value over gene-level features for perturbation-target prediction. — Kendiukhov I., BMC genomics, 2026
Related models & tools相关模型与工具
- VC0019 Predicting cellular responses to perturbation across diverse contexts with State. — Adduri AK et al., Cell, 2026
- VC0033 · preprint · 预印本 X-Cell: Scaling Causal Perturbation Prediction Across Diverse Cellular Contexts via Diffusion Language Models — Wang C et al., bioRxiv, 2026
Perturbation & validation evidence扰动与验证证据
- VC0002 Virtual Cell Challenge: Toward a Turing test for the virtual cell. — Roohani YH et al., Cell, 2025
- VC0058 Virtual Cell Challenge 2026: Benchmarking zero-shot generalization across cellular contexts. — Arc Virtual Cell Initiative Team, Goodarzi H., Cell, 2026
- VC0053 · preprint · 预印本 Signal, Bounds, and Baselines: Principles for Evaluating Virtual Cell Perturbation Models — Vollenweider M, Bühlmann P., bioRxiv, 2026
- VC0054 · preprint · 预印本 An Empirical Comparison of Virtual Cell Models: Perturbation Prediction, Representation, and the Baseline Gap — Denvis O., Research Square, 2026
- VC0077 Generative AI for spatial tumor growth on MRI: a proof-of-principle study in pediatric diffuse midline glioma. — Laslo D et al., BMC medicine, 2026
- VC0078 Biological domain shift and statistical nesting concerns in generative AI-based spatial tumor growth prediction for pediatric diffuse midline glioma. — Li W, Sun X, Pei S., BMC medicine, 2026
13 library items cited on this page; 3 are preprints (not peer reviewed).本页引用文献库条目 13 条，其中预印本（未经同行评审）3 条。
What remains uncertain尚不确定之处
Benchmarks are mostly on cell lines; there is no GBM-specific benchmark in the library.基准多基于细胞系；文献库中尚无 GBM 特异基准。
Some newer evidence on the baseline gap is in preprints.关于基线差距的部分新证据来自预印本。 VC0053 · preprint · 预印本 VC0054 · preprint · 预印本
For research and education only. Not medical advice and not a recommendation of any diagnosis, treatment or product. Patients should consult their own physicians.仅供科研与教育参考，不构成医疗建议，也不构成对任何诊疗或产品的推荐。患者请咨询主治医生。