Canon enables causal inference of downstream genes in single-cell CRISPR studies via instrumental variable analysis

• 成果类型: Article
• 署名作者: Wu, Peijun; Zhou, Xiang
• 署名单位: University of Michigan System; University of Michigan; University of Michigan System; University of Michigan; Yale University
• 刊物名称: PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
• ISSN/ISSBN: 0027-8424; 1091-6490
• DOI: 10.1073/pnas.2525359123
• 发表日期: 2026-08-18
• 页码: e2525359123
• 关键词: single-cell CRISPR; genomics; instrumental variable analysis; statistical and computational method; causal relationship discovery; mendelian randomization; identification; circuits; genome

摘要

A critical analytical task in sc-CRISPR screening is identifying downstream genes influenced by perturbed target genes. Existing methods for this task primarily rely on traditional association-based analyses, which not only fall short in establishing causal relationships between genes but also suffer from high false positive rates and limited statistical power. To overcome these limitations, we introduce a causal inference-based framework that leverages the perturbation status of gRNAs in single cells as instrumental variables (IVs) to infer causal gene relationship via IV analysis. Building upon this framework, we further present Canon, a one-sample IV analysis method specifically tailored to systematically identify genes that are potentially causally influenced by perturbed target genes across diverse sc-CRISPR platforms. Canon ensures robust type I error control while maintaining high statistical power. We evaluated its performance through comprehensive simulations and real data applications. The gene-gene relationships identified by Canon provide valuable insights into the causal gene regulatory network, uncovering candidate therapeutic targets with potential relevance for cancer biology and demonstrating the transformative potential of sc-CRISPR screening to resolve causal regulatory networks at an unprecedented scale.