Abstract
In order to increase statistical power for learning a causal network, data are often pooled from multiple observational and interventional experiments. However, if the direct effects of interventions are uncertain, multi-experiment data pooling can result in false causal discoveries. We present a new method, “Learn and Vote,” for inferring causal interactions from multi-experiment datasets. In our method, experiment-specific networks are learned from the data and then combined by weighted averaging to construct a consensus network. Through empirical studies on synthetic and real-world datasets, we found that for most of the larger-sized network datasets that we analyzed, our method is more accurate than state-of-the-art network inference approaches.
Highlights
Causal modeling is an important analytical paradigm in action planning, predictive applications, research, and medical diagnosis [1, 2]
Based on prior studies suggesting that incorporating data from interventional experiments improves network inference, we re-analyzed a small subset of the Sachs et al [16] biological cell signaling dataset using their published inference approach, two times
We report a new approach, “Learn and Vote,” for learning a causal network structure from multiple datasets generated from different experiments, including the case of hybrid observational-interventional datasets
Summary
Causal modeling is an important analytical paradigm in action planning, predictive applications, research, and medical diagnosis [1, 2]. Vj, where Vi and Vj represent observable entities and the direction of the arrow denotes that the state of Vi influences the state of Vj. Causal models can be inferred from passive observational measurements (“seeing”) and by measurements collected after performing external interventions (“doing”) on the states of the domain entities. Observational measurements [3] are more straightforward to obtain than interventional measurements, and observational datasets are frequently used for causal inference. Vk, are Markov equivalent—each encodes the conditional independence statement Vi ╨ Vk|Vj. For example, the three causal models Vi ! Vk, are Markov equivalent—each encodes the conditional independence statement Vi ╨ Vk|Vj This ambiguity can in principle be resolved by incorporating measurements obtained from interventional experiments in which specific entities are targeted with perturbations.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
