{"id":"perturbation-prediction","type":"method","name_en":"Perturbation response prediction","name_zh":"扰动反应预测","aliases":["perturbation prediction","perturbation effect prediction"],"description":{"en":"Predicting transcriptional responses to genetic or drug perturbations, as done by GEARS, CPA and State. Independent benchmarks report that deep-learning models do not yet reliably outperform simple linear baselines.","zh":"预测基因或药物扰动后的转录反应，如 GEARS、CPA 与 State 所做。独立基准研究报告深度学习模型尚不能稳定超越简单线性基线。"},"externalIds":{},"graphNames":["perturbation prediction"],"mentions":{"items":["VC0015","VC0016","VC0018","VC0033","VC0048","VC0049","VC0051","VC0052","VC0054","VC0055","VC0129","VC0138"],"topics":["model-virtual-cell","validation-benchmarks"]},"evidenceRefs":["VC0013","VC0014","VC0019","VC0048","VC0049"],"reviewStatus":"draft","version":"0.1.0","page_url":"https://gliostate.com/entities/perturbation-prediction/","page_url_zh":"https://gliostate.com/zh/entities/perturbation-prediction/","api_url":"https://gliostate.com/api/entities/perturbation-prediction.json","external_links":[],"source_file":"https://gliostate.com/data/entities/perturbation-prediction.json","evidence_items":[{"id":"VC0013","title":"Predicting transcriptional outcomes of novel multigene perturbations with GEARS.","year":"2023","page_url":"https://gliostate.com/library/VC0013/","api_url":"https://gliostate.com/api/items/VC0013.json"},{"id":"VC0014","title":"Predicting cellular responses to complex perturbations in high-throughput screens.","year":"2023","page_url":"https://gliostate.com/library/VC0014/","api_url":"https://gliostate.com/api/items/VC0014.json"},{"id":"VC0019","title":"Predicting cellular responses to perturbation across diverse contexts with State.","year":"2026","page_url":"https://gliostate.com/library/VC0019/","api_url":"https://gliostate.com/api/items/VC0019.json"},{"id":"VC0048","title":"Deep-learning-based gene perturbation effect prediction does not yet outperform simple linear baselines.","year":"2025","page_url":"https://gliostate.com/library/VC0048/","api_url":"https://gliostate.com/api/items/VC0048.json"},{"id":"VC0049","title":"Benchmarking foundation cell models for post-perturbation RNA-seq prediction.","year":"2025","page_url":"https://gliostate.com/library/VC0049/","api_url":"https://gliostate.com/api/items/VC0049.json"}],"mentioned_in_items":[{"id":"VC0015","title":"scGen predicts single-cell perturbation responses.","year":"2019","page_url":"https://gliostate.com/library/VC0015/","api_url":"https://gliostate.com/api/items/VC0015.json"},{"id":"VC0016","title":"Learning single-cell perturbation responses using neural optimal transport.","year":"2023","page_url":"https://gliostate.com/library/VC0016/","api_url":"https://gliostate.com/api/items/VC0016.json"},{"id":"VC0018","title":"TxPert: using multiple knowledge graphs for prediction of transcriptomic perturbation effects.","year":"2026","page_url":"https://gliostate.com/library/VC0018/","api_url":"https://gliostate.com/api/items/VC0018.json"},{"id":"VC0033","title":"X-Cell: Scaling Causal Perturbation Prediction Across Diverse Cellular Contexts via Diffusion Language Models","year":"2026","page_url":"https://gliostate.com/library/VC0033/","api_url":"https://gliostate.com/api/items/VC0033.json"},{"id":"VC0048","title":"Deep-learning-based gene perturbation effect prediction does not yet outperform simple linear baselines.","year":"2025","page_url":"https://gliostate.com/library/VC0048/","api_url":"https://gliostate.com/api/items/VC0048.json"},{"id":"VC0049","title":"Benchmarking foundation cell models for post-perturbation RNA-seq prediction.","year":"2025","page_url":"https://gliostate.com/library/VC0049/","api_url":"https://gliostate.com/api/items/VC0049.json"},{"id":"VC0051","title":"PertEval-scFM: Benchmarking Single-Cell Foundation Models for Perturbation Effect Prediction","year":"2024","page_url":"https://gliostate.com/library/VC0051/","api_url":"https://gliostate.com/api/items/VC0051.json"},{"id":"VC0052","title":"Systema: a framework for evaluating genetic perturbation response prediction beyond systematic variation.","year":"2025","page_url":"https://gliostate.com/library/VC0052/","api_url":"https://gliostate.com/api/items/VC0052.json"},{"id":"VC0054","title":"An Empirical Comparison of Virtual Cell Models: Perturbation Prediction, Representation, and the Baseline Gap","year":"2026","page_url":"https://gliostate.com/library/VC0054/","api_url":"https://gliostate.com/api/items/VC0054.json"},{"id":"VC0055","title":"Multimodal Benchmarking of Foundation Model Representations for Cellular Perturbation Response Prediction","year":"2025","page_url":"https://gliostate.com/library/VC0055/","api_url":"https://gliostate.com/api/items/VC0055.json"},{"id":"VC0129","title":"DELPHAI, AI Agent for Predicting Drug Response and Resistance","year":"2025","page_url":"https://gliostate.com/library/VC0129/","api_url":"https://gliostate.com/api/items/VC0129.json"},{"id":"VC0138","title":"PerturbLDM: conditional latent diffusion for modelling single-cell perturbation responses","year":"2026","page_url":"https://gliostate.com/library/VC0138/","api_url":"https://gliostate.com/api/items/VC0138.json"}],"mentioned_in_topics":[{"id":"model-virtual-cell","title_en":"AI virtual cell models","title_zh":"AI 虚拟细胞模型","url":"https://gliostate.com/models/virtual-cell/","api_url":"https://gliostate.com/api/topics/model-virtual-cell.json"},{"id":"validation-benchmarks","title_en":"Validating virtual cells: benchmarks, baselines and critiques","title_zh":"虚拟细胞的验证：基准、基线与批评","url":"https://gliostate.com/validation/benchmarks-and-baselines/","api_url":"https://gliostate.com/api/topics/validation-benchmarks.json"}],"relations":[{"subject":"perturbation-prediction","subject_name_en":"Perturbation response prediction","subject_name_zh":"扰动反应预测","predicate":"benchmarked_on","object":"linear-baseline","object_name_en":"Simple linear baseline","object_name_zh":"简单线性基线","confidence":"high","evidence":[{"id":"VC0048","role":"benchmark","title":"Deep-learning-based gene perturbation effect prediction does not yet outperform simple linear baselines.","page_url":"https://gliostate.com/library/VC0048/","api_url":"https://gliostate.com/api/items/VC0048.json"},{"id":"VC0049","role":"benchmark","title":"Benchmarking foundation cell models for post-perturbation RNA-seq prediction.","page_url":"https://gliostate.com/library/VC0049/","api_url":"https://gliostate.com/api/items/VC0049.json"}],"note":{"en":"Deep models do not yet reliably outperform simple linear baselines.","zh":"深度模型尚不能稳定超越简单线性基线。"},"sources":[{"kind":"topic","id":"validation-benchmarks","url":"https://gliostate.com/validation/benchmarks-and-baselines/","api_url":"https://gliostate.com/api/topics/validation-benchmarks.json"}]},{"subject":"gears","subject_name_en":"GEARS","subject_name_zh":"GEARS","predicate":"associated_with","object":"perturbation-prediction","object_name_en":"Perturbation response prediction","object_name_zh":"扰动反应预测","confidence":"high","evidence":[{"id":"VC0013","role":"supports","title":"Predicting transcriptional outcomes of novel multigene perturbations with GEARS.","page_url":"https://gliostate.com/library/VC0013/","api_url":"https://gliostate.com/api/items/VC0013.json"}],"note":null,"sources":[{"kind":"topic","id":"model-virtual-cell","url":"https://gliostate.com/models/virtual-cell/","api_url":"https://gliostate.com/api/topics/model-virtual-cell.json"}]},{"subject":"cpa","subject_name_en":"CPA (compositional perturbation autoencoder)","subject_name_zh":"CPA（组合扰动自编码器）","predicate":"associated_with","object":"perturbation-prediction","object_name_en":"Perturbation response prediction","object_name_zh":"扰动反应预测","confidence":"high","evidence":[{"id":"VC0014","role":"supports","title":"Predicting cellular responses to complex perturbations in high-throughput screens.","page_url":"https://gliostate.com/library/VC0014/","api_url":"https://gliostate.com/api/items/VC0014.json"}],"note":null,"sources":[{"kind":"topic","id":"model-virtual-cell","url":"https://gliostate.com/models/virtual-cell/","api_url":"https://gliostate.com/api/topics/model-virtual-cell.json"}]},{"subject":"state","subject_name_en":"State (Arc Institute)","subject_name_zh":"State（Arc 研究所）","predicate":"associated_with","object":"perturbation-prediction","object_name_en":"Perturbation response prediction","object_name_zh":"扰动反应预测","confidence":"high","evidence":[{"id":"VC0019","role":"supports","title":"Predicting cellular responses to perturbation across diverse contexts with State.","page_url":"https://gliostate.com/library/VC0019/","api_url":"https://gliostate.com/api/items/VC0019.json"}],"note":null,"sources":[{"kind":"topic","id":"model-virtual-cell","url":"https://gliostate.com/models/virtual-cell/","api_url":"https://gliostate.com/api/topics/model-virtual-cell.json"}]},{"subject":"systematic-variation","subject_name_en":"Systematic variation (evaluation concept)","subject_name_zh":"系统性差异（评估概念）","predicate":"associated_with","object":"perturbation-prediction","object_name_en":"Perturbation response prediction","object_name_zh":"扰动反应预测","confidence":"high","evidence":[{"id":"VC0052","role":"benchmark","title":"Systema: a framework for evaluating genetic perturbation response prediction beyond systematic variation.","page_url":"https://gliostate.com/library/VC0052/","api_url":"https://gliostate.com/api/items/VC0052.json"}],"note":{"en":"Systematic variation can lead to overestimated predictor performance.","zh":"系统性差异可能导致预测器性能被高估。"},"sources":[{"kind":"topic","id":"validation-benchmarks","url":"https://gliostate.com/validation/benchmarks-and-baselines/","api_url":"https://gliostate.com/api/topics/validation-benchmarks.json"}]},{"subject":"x-cell","subject_name_en":"X-Cell (Xaira)","subject_name_zh":"X-Cell（Xaira）","predicate":"associated_with","object":"perturbation-prediction","object_name_en":"Perturbation response prediction","object_name_zh":"扰动反应预测","confidence":"medium","evidence":[{"id":"VC0033","role":"supports","title":"X-Cell: Scaling Causal Perturbation Prediction Across Diverse Cellular Contexts via Diffusion Language Models","page_url":"https://gliostate.com/library/VC0033/","api_url":"https://gliostate.com/api/items/VC0033.json"}],"note":{"en":"Preprint.","zh":"预印本。"},"sources":[{"kind":"topic","id":"model-virtual-cell","url":"https://gliostate.com/models/virtual-cell/","api_url":"https://gliostate.com/api/topics/model-virtual-cell.json"}]}],"related_entities":[{"id":"linear-baseline","type":"method","name_en":"Simple linear baseline","name_zh":"简单线性基线","page_url":"https://gliostate.com/entities/linear-baseline/","page_url_zh":"https://gliostate.com/zh/entities/linear-baseline/","api_url":"https://gliostate.com/api/entities/linear-baseline.json","via":"relation"},{"id":"gears","type":"model","name_en":"GEARS","name_zh":"GEARS","page_url":"https://gliostate.com/entities/gears/","page_url_zh":"https://gliostate.com/zh/entities/gears/","api_url":"https://gliostate.com/api/entities/gears.json","via":"relation"},{"id":"cpa","type":"model","name_en":"CPA (compositional perturbation autoencoder)","name_zh":"CPA（组合扰动自编码器）","page_url":"https://gliostate.com/entities/cpa/","page_url_zh":"https://gliostate.com/zh/entities/cpa/","api_url":"https://gliostate.com/api/entities/cpa.json","via":"relation"},{"id":"state","type":"model","name_en":"State (Arc Institute)","name_zh":"State（Arc 研究所）","page_url":"https://gliostate.com/entities/state/","page_url_zh":"https://gliostate.com/zh/entities/state/","api_url":"https://gliostate.com/api/entities/state.json","via":"relation"},{"id":"systematic-variation","type":"method","name_en":"Systematic variation (evaluation concept)","name_zh":"系统性差异（评估概念）","page_url":"https://gliostate.com/entities/systematic-variation/","page_url_zh":"https://gliostate.com/zh/entities/systematic-variation/","api_url":"https://gliostate.com/api/entities/systematic-variation.json","via":"relation"},{"id":"x-cell","type":"model","name_en":"X-Cell (Xaira)","name_zh":"X-Cell（Xaira）","page_url":"https://gliostate.com/entities/x-cell/","page_url_zh":"https://gliostate.com/zh/entities/x-cell/","api_url":"https://gliostate.com/api/entities/x-cell.json","via":"relation"},{"id":"foundation-model","type":"method","name_en":"Single-cell foundation model","name_zh":"单细胞基础模型","page_url":"https://gliostate.com/entities/foundation-model/","page_url_zh":"https://gliostate.com/zh/entities/foundation-model/","api_url":"https://gliostate.com/api/entities/foundation-model.json","via":"co-mention"},{"id":"zero-shot-evaluation","type":"method","name_en":"Zero-shot evaluation","name_zh":"零样本评估","page_url":"https://gliostate.com/entities/zero-shot-evaluation/","page_url_zh":"https://gliostate.com/zh/entities/zero-shot-evaluation/","api_url":"https://gliostate.com/api/entities/zero-shot-evaluation.json","via":"co-mention"}]}