Causal Effect Estimation via Doubly Robust One-Step Estimators and TMLE in Graphical Models with Unmeasured Variables // Provides doubly robust one-step and targeted maximum likelihood (TMLE) estimators for average causal effects in acyclic directed mixed graphs (ADMGs) with unmeasured variables. Automatically determines whether the treatment effect is identified via backdoor adjustment or the extended front-door functional, and dispatches to the appropriate estimator. Supports incorporation of machine learning algorithms via 'SuperLearner' and cross-fitting for nuisance estimation. Methods are described in Guo and Nabi (2024) doi:10.48550/arXiv.2409.03962.