New methods for separating causes from effects in genomics data

Abstract

BackgroundThe discovery of molecular pathways is a challenging problem and its solution relies on the identification of causal molecular interactions in genomics data. Causal molecular interactions can be discovered using randomized experiments; however such experiments are often costly, infeasible, or unethical. Fortunately, algorithms that infer causal interactions from observational data have been in development for decades, predominantly in the quantitative sciences, and many of them have recently been applied to genomics data. While these algorithms can infer unoriented causal interactions between involved molecular variables (i.e., without specifying which one is the cause and which one is the effect), causally orienting all inferred molecular interactions was assumed to be an unsolvable problem until recently. In this work, we use transcription factor-target gene regulatory interactions in three different organisms to evaluate a new family of methods that, given observational data for just two causally related variables, can determine which one is the cause and which one is the effect.ResultsWe have found that a particular family of causal orientation methods (IGCI Gaussian) is often able to accurately infer directionality of causal interactions, and that these methods usually outperform other causal orientation techniques. We also introduced a novel ensemble technique for causal orientation that combines decisions of individual causal orientation methods. The ensemble method was found to be more accurate than any best individual causal orientation method in the tested data.ConclusionsThis work represents a first step towards establishing context for practical use of causal orientation methods in the genomics domain. We have found that some causal orientation methodologies yield accurate predictions of causal orientation in genomics data, and we have improved on this capability with a novel ensemble method. Our results suggest that these methods have the potential to facilitate reconstruction of molecular pathways by minimizing the number of required randomized experiments to find causal directionality and by avoiding experiments that are infeasible and/or unethical.

Plasma levels of interleukin-1 receptor antagonist (IL1Ra) predict radiographic progression of symptomatic knee osteoarthritis

“…As elegantly observed by Bruyere et al 39 , “Decisions on the selection of appropriate therapy can be made considering clinical presentation, underlying pathophysiology and stage of OA disease. Furthermore, identification of these patient profiles and phenotype of OA may lead to more personalized healthcare and more targeted treatment for osteoarthritis.” We investigated here the application of cutting-edge causal graph techniques that have recently been shown to have significant de novo mechanism and pathway discovery capabilities 59–61 ; this approach has provided evidence that IL1Ra – in an as-yet-undefined interaction with BMI – predicts radiographic medial JSN progression, and our ongoing research in a larger OAI cohort may provide deeper understanding of the role of plasma IL1Ra in pathophysiology of OA.…”

Analysis of cause-effect inference by comparing regression errors

“…In the following, we use the term 'causal inference' to refer to the identification of the true causal direction. A possible application is the discovery of molecular pathways, which relies on the identification of causal molecular interactions in genomics data (Statnikov et al, 2012). Other examples in biomedicine where observational data can be used for causal discovery are discussed in the work by Ma & Statnikov (2017).…”

Explainable and Interpretable Models in Computer Vision and Machine Learning

“…Other causal inference methods (Sgouritsa et al, 2015) are based on the idea that if X → Y , the marginal probability distribution of the cause P(X) is independent of the causal mechanism P(Y |X), hence estimating P(Y |X) from P(X) should hardly be possible, while estimating P(X|Y ) based on P(Y ) may be possible. The reader is referred to Statnikov et al (2012) and Mooij et al (2016) for a thorough review and benchmark of the pairwise methods in the bivariate case.…”