Local Causal Structure Learning in the Presence of Latent Variables and Selection Bias

2026-07-22

AI Summary

This work proposes LoCaLS, the first complete and sound target-oriented local causal discovery algorithm for observational data involving latent variables and selection bias. LoCaLS accurately identifies direct causes and effects of a target variable without requiring reconstruction of the full global causal graph. By theoretically linking local observational distributions to global causal structures and integrating causal graph theory, conditional independence tests, and an efficient local search strategy, the method substantially reduces computational overhead. Empirical evaluations demonstrate that LoCaLS consistently outperforms existing local approaches on both synthetic and real-world datasets, achieves markedly higher computational efficiency than state-of-the-art global algorithms, and successfully uncovers biologically meaningful causal relationships in gene expression data.

Abstract

Discovering the direct causes and effects of a target variable from observational data is a fundamental problem in causal discovery, with broad applications in domains such as gene regulatory analysis and biomedical research. Existing causal discovery methods either learn a global causal structure, which incurs substantial computational cost, or assume the absence of latent variables and selection bias, assumptions that are often violated in real-world settings. Motivated by these challenges, we study local causal structure learning in the presence of latent variables and selection bias. Specifically, we first characterize a local region that enables target-specific causal discovery without recovering the entire global structure. We then establish a theoretical bridge between causal information learned from the observed distribution induced on this local region and the corresponding information in the global causal structure. Building on these foundations, we propose LoCaLS, a local causal structure learning algorithm that is sound and complete under standard assumptions and identifies the same direct causes and effects of a target variable as those identifiable by global causal discovery methods, while allowing for latent variables and selection bias. Extensive experiments on random and real-world structures demonstrate that the proposed method consistently achieves higher structural accuracy than existing local methods while requiring substantially less computational effort than state-of-the-art global methods. Furthermore, applications to two real-world gene expression datasets reveal biologically plausible target-specific causal structures, demonstrating its practical applicability in large-scale biological data analysis.

Problem

• causal discovery
• latent variables
• selection bias
• local causal structure
• target variable

Innovation

• local causal discovery
• latent variables
• selection bias
• causal structure learning
• LoCaLS