Permutation-based Causal Inference Algorithms with Interventions

About

Learning directed acyclic graphs using both observational and interventional data is now a fundamentally important problem due to recent technological developments in genomics that generate such single-cell gene expression data at a very large scale. In order to utilize this data for learning gene regulatory networks, efficient and reliable causal inference algorithms are needed that can make use of both observational and interventional data. In this paper, we present two algorithms of this type and prove that both are consistent under the faithfulness assumption. These algorithms are interventional adaptations of the Greedy SP algorithm and are the first algorithms using both observational and interventional data with consistency guarantees. Moreover, these algorithms have the advantage that they are nonparametric, which makes them useful also for analyzing non-Gaussian data. In this paper, we present these two algorithms and their consistency guarantees, and we analyze their performance on simulated data, protein signaling data, and single-cell gene expression data.

Yuhao Wang, Liam Solus, Karren Dai Yang, Caroline Uhler · 2017

Related benchmarks

Task	Dataset	Result
Causal Discovery	Synthetic Data	Runtime0	57
Causal Discovery	Semantic Causal Environment observation-only	F1 Score66.4	15
Causal Discovery	Synthetic Data Observation-only (1000 samples)	Rank7.2	15
Causal Discovery	Causal Chambers Light-tunnel real data results	Precision0	14
Causal Discovery	SACHS p = 11, s = 20, n = 100 (real flow cytometry)	F1 Score24	13
Causal Discovery	bnlearn Earthquake standard (test)	SHD0	10
Causal Discovery	bnlearn Survey standard (test)	SHD1	10
Causal Discovery	Nonlinear SCM 1.0 (synthetic)	F1 (NL1)52.5	9
Causal Discovery	Synthetic Linear SCMs Fork, Chain, V-str, Diamond, Collider (test)	Fork Score0.472	9
Causal Discovery	CausalDynamics Lorenz	AUROC66	9