Boosting Local Causal Discovery in High-Dimensional Expression Data

TL;DR

This paper enhances local causal discovery methods for high-dimensional gene expression data by developing practical estimators that are simpler and more efficient, achieving accuracy close to state-of-the-art techniques.

Contribution

It introduces tailored estimators for LCD in high-dimensional settings, incorporating preselection and statistical tests to improve efficiency and accuracy.

Findings

LCD estimator approaches ICP accuracy

Method is simpler and more computationally efficient

Effective in large-scale gene expression data

Abstract

We study the performance of Local Causal Discovery (LCD), a simple and efficient constraint-based method for causal discovery, in predicting causal effects in large-scale gene expression data. We construct practical estimators specific to the high-dimensional regime. Inspired by the ICP algorithm, we use an optional preselection method and two different statistical tests. Empirically, the resulting LCD estimator is seen to closely approach the accuracy of ICP, the state-of-the-art method, while it is algorithmically simpler and computationally more efficient.