Evaluation and characterisation of differential co-expression analysis methods

Abstract

Elucidation of regulatory networks is a key aim in systems biology, including identification of regulatory mechanisms specific to a given biological context. This has motivated the move from co-expression to differential co-expression analysis and numerous methods have been developed subsequently to address this problem. Evaluation of methods and interpretation of resulting networks has been hindered by the lack of known context-specific regulatory interactions. We develop a simulator based on dynamical systems modelling capable of simulating differential co-expression patterns. We then used this to integrated the simulator into an evaluation framework to benchmark and characterise the performance of inference methods. Defining three different levels of “true” networks for each simulation, we showed that accurate inference of causation from transcriptomic data alone was difficult for all methods, compared to inference of associations. We show that a z-score based method had the best general performance. Further, analysis of simulation parameters revealed topological properties that explained the performance of methods. Analysis of inferred networks showed how hub nodes were more likely to be differentially regulated targets than transcription factors regulating alternate programs. This led up to propose an interpretation of the inferred differential network that helps us reconstruct a putative causative network in breast cancer. Application to a breast cancer dataset revealed differential regulation of immune processes dependent on estrogen receptor status. The potential of differential co-expression analysis remains largely unexplored due to difficulties in interpreting results, and we have attempted to address some of the limiting factors. Applications of methods are not limited to co-expression and may be applied to associations in general.

Dharmesh D Bhuva

Senior post-doctoral researcher at SAiGENCI

My research interests include cancer systems biology, spatial statistics and computational biology.