Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.19343 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910234999521280 |
|---|---|
| author | Jiang, Wenkang Liu, Yuhang Cai, Yichao Gao, Erdun Dong, Jiayi Abbasnejad, Ehsan Yao, Lina Shi, Javen Qinfeng |
| author_facet | Jiang, Wenkang Liu, Yuhang Cai, Yichao Gao, Erdun Dong, Jiayi Abbasnejad, Ehsan Yao, Lina Shi, Javen Qinfeng |
| contents | Single-cell perturbation modeling is fundamental for understanding and predicting cellular responses to genetic perturbations. However, existing approaches, from causal representation learning to foundation models, often struggle with an overlooked challenge: gene expression is dominated by perturbation-invariant information, while perturbation-specific signals are intrinsically sparse. As a result, learned representations either entangle invariant and perturbation-specific information, leading to spurious and non-generalizable predictors, or suppress perturbation-specific signals altogether, rendering them ineffective for prediction. To address this, we propose PerturbedVAE, a general framework designed to resolve this signal imbalance. The framework explicitly separates perturbation-specific information from dominant invariant structure and recovers causal representations to effectively utilize such information for prediction. We further provide an identifiability analysis that characterizes the conditions under which sparse perturbation effects can be reliably recovered, thereby clarifying how the framework can be concretely specified under such conditions. Empirically, PerturbedVAE achieves state-of-the-art performance on a widely used benchmark across multiple evaluation settings, yielding significant gains on out-of-distribution combinatorial predictions and uncovering interpretable perturbation-response programs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_19343 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | What Makes a Representation Good for Single-Cell Perturbation Prediction? Jiang, Wenkang Liu, Yuhang Cai, Yichao Gao, Erdun Dong, Jiayi Abbasnejad, Ehsan Yao, Lina Shi, Javen Qinfeng Machine Learning Single-cell perturbation modeling is fundamental for understanding and predicting cellular responses to genetic perturbations. However, existing approaches, from causal representation learning to foundation models, often struggle with an overlooked challenge: gene expression is dominated by perturbation-invariant information, while perturbation-specific signals are intrinsically sparse. As a result, learned representations either entangle invariant and perturbation-specific information, leading to spurious and non-generalizable predictors, or suppress perturbation-specific signals altogether, rendering them ineffective for prediction. To address this, we propose PerturbedVAE, a general framework designed to resolve this signal imbalance. The framework explicitly separates perturbation-specific information from dominant invariant structure and recovers causal representations to effectively utilize such information for prediction. We further provide an identifiability analysis that characterizes the conditions under which sparse perturbation effects can be reliably recovered, thereby clarifying how the framework can be concretely specified under such conditions. Empirically, PerturbedVAE achieves state-of-the-art performance on a widely used benchmark across multiple evaluation settings, yielding significant gains on out-of-distribution combinatorial predictions and uncovering interpretable perturbation-response programs. |
| title | What Makes a Representation Good for Single-Cell Perturbation Prediction? |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2605.19343 |