Latent Causal Diffusions for Single-Cell Perturbation Modeling

2026-01-20

arXiv.org

AI Summary

Existing methods struggle to accurately model single-cell transcriptomic responses to perturbations, often conflating measurement noise with true biological signals and failing to uncover the underlying causal regulatory architecture. To address these limitations, this work proposes the LCD-CLIPR framework, which formulates single-cell gene expression as a noise-perturbed stationary diffusion process. By integrating a Latent Causal Diffusion (LCD) model with a Causal Linearization for Perturbation Response (CLIPR) approach, the method embeds stochastic differential equations and causal identification theory into generative modeling. This represents the first successful integration of generative diffusion models with causal inference, enabling accurate prediction of expression distributions under unseen perturbation combinations. Evaluated on both simulated and real single-cell RNA-seq perturbation datasets, LCD-CLIPR substantially outperforms existing methods, recovering genome-wide direct causal regulatory relationships and revealing functional modules and regulatory mechanisms undetectable by conventional differential expression analysis.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Computer Vision: Diffusion Models for Vision

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Perturbation screens hold the potential to systematically map regulatory processes at single-cell resolution, yet modeling and predicting transcriptome-wide responses to perturbations remains a major computational challenge. Existing methods often underperform simple baselines, fail to disentangle measurement noise from biological signal, and provide limited insight into the causal structure governing cellular responses. Here, we present the latent causal diffusion (LCD), a generative model that frames single-cell gene expression as a stationary diffusion process observed under measurement noise. LCD outperforms established approaches in predicting the distributional shifts of unseen perturbation combinations in single-cell RNA-sequencing screens while simultaneously learning a mechanistic dynamical system of gene regulation. To interpret these learned dynamics, we develop an approach we call causal linearization via perturbation responses (CLIPR), which yields an approximation of the direct causal effects between all genes modeled by the diffusion. CLIPR provably identifies causal effects under a linear drift assumption and recovers causal structure in both simulated systems and a genome-wide perturbation screen, where it clusters genes into coherent functional modules and resolves causal relationships that standard differential expression analysis cannot. The LCD-CLIPR framework bridges generative modeling with causal inference to predict unseen perturbation effects and map the underlying regulatory mechanisms of the transcriptome.

Problem

• single-cell perturbation
• causal inference
• transcriptome modeling
• gene regulation
• distributional shift

Innovation

• latent causal diffusion
• causal inference
• single-cell perturbation
• generative modeling
• gene regulatory dynamics