dce

The Differential Causal Effects (dce) is a novel computational approach designed to identify dysregulated signaling pathways in cancer cells by comparing them to normal cells within the statistical causality framework. This method focuses on detecting individual edges in a signaling pathway that exhibit dysregulation in cancer cells while considering and adjusting for confounding factors. By accounting for confounding, dce is more resilient to the influence of technical artifacts, increasing the likelihood of identifying true biological signals.
The dce approach is extended to handle unobserved dense confounding, where latent variables, such as batch effects or cell cycle states, affect many covariates. Through validation on synthetic datasets, CRISPR knockout screens, and a GTEx dataset, dce outperforms competing methods and demonstrates effective adjustment for latent confounding.