Proxy-Adjusted Causal Discovery from Targeted Interventions
Abstract
Randomized perturbations can reveal downstream responses without identifying which causal relationships are direct. Conditioning on intermediate responses may induce associations through unmeasured common causes, even when the source is randomized. We introduce Proxy-Adjusted Balanced Masking (PABM), a nonparametric framework for recovering directed acyclic graphs from targeted interventions, response data, and recorded proxies. PABM uses valid target-specific interventions to identify candidate ancestors. It then compares conditional target distributions with and without a candidate response and, when supported by the design, its intervention variable, adjusting for other ancestors and proxies. Each comparison uses matched observations, adjustment variables, and fitting procedures. Identification requires every included nonparent comparison to have zero conditional gain and each parent to have positive gain in at least one comparison. We establish population identification, finite-sample recovery conditions that accommodate incomplete adjustment and estimation error, and familywise error control under valid held-out p-values. Continuous-response simulations show favorable graph recovery relative to specified comparator pipelines; K562-calibrated count simulations reveal a selection--ranking tradeoff. Proxy ablations assess sensitivity to recorded information. An analysis of K562 Perturb-seq data illustrates descriptive candidate-network construction; proxy adequacy for unmeasured biological variation remains unresolved.
Related
- Statistical Inference for Causal Discovery under Selection and Latent Variables via Single-Target Interventions
- Anytime-Valid Inference in Linear Models with Applications to Regression-Adjusted Causal Inference
- Proximal causal inference through cross-proxy balancing
- FOCUS: Fixed-Confidence Online Causal Learning Using Sequential Adaptive Interventions
- Collaborative representations for targeted causal inference under outcome selection
BibTeX
@misc{chen2026proxyadjusted,
title = {{Proxy-Adjusted Causal Discovery from Targeted Interventions}},
author = {Li Chen and Wei Pan and Xiaotong Shen},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.23897},
eprint = {2609.23897},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.23897},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Proxy-Adjusted Causal Discovery from Targeted Interventions
AU - Li Chen
AU - Wei Pan
AU - Xiaotong Shen
PY - 2026
DA - 2026/09/20
PB - arXiv
DO - 10.48550/arxiv.2609.23897
UR - https://arxiv.org/abs/2609.23897
AB - Randomized perturbations can reveal downstream responses without identifying which causal relationships are direct. Conditioning on intermediate responses may induce associations through unmeasured common causes, even when the source is randomized. We introduce Proxy-Adjusted Balanced Masking (PABM), a nonparametric framework for recovering directed acyclic graphs from targeted interventions, response data, and recorded proxies. PABM uses valid target-specific interventions to identify candidate ancestors. It then compares conditional target distributions with and without a candidate response and, when supported by the design, its intervention variable, adjusting for other ancestors and proxies. Each comparison uses matched observations, adjustment variables, and fitting procedures. Identification requires every included nonparent comparison to have zero conditional gain and each parent to have positive gain in at least one comparison. We establish population identification, finite-sample recovery conditions that accommodate incomplete adjustment and estimation error, and familywise error control under valid held-out p-values. Continuous-response simulations show favorable graph recovery relative to specified comparator pipelines; K562-calibrated count simulations reveal a selection--ranking tradeoff. Proxy ablations assess sensitivity to recorded information. An analysis of K562 Perturb-seq data illustrates descriptive candidate-network construction; proxy adequacy for unmeasured biological variation remains unresolved.
ER -