Examining Big Data Using Information Flow Analysis - Application To The Functional Characterization Of Plant Derived Therapeutics

Biological complexity results from redundancy among systems and processes which allow growth, reproduction and responses to the environment. Understanding their connectivity requires integrative, multi‐scale analytical approaches. However, this integration has been frustrated by heterogeneity in the sorts of data sets available, i.e., types of molecules, scales of measurement, types of model systems. Here, we describe an approach called Information Flow Analysis which works through alignment of cause‐effect spectra. These spectra link a defined vocabulary causes (for instance drugs) with a defined vocabulary of effects (for instance physiological responses). An individual spectrum might be an array row with one drug element and a list of effect elements. Each effect element would record the number of co‐citations for a drug and a particular physiological effect, occurring over the sum of all biological abstracts, for instance. Elements could be full or empty. As these elements are dimensionless, they eliminate data heterogeneity problems and the associated spectra are thus directly comparable. These strings of numbers (spectra) can be hierarchically clustered, with effects clustered by similarity in causes and causes clustered by similarity in effects. This type of analysis is scalable, recapitulates known biology, and elicits new relationships. Here, we apply this approach for the functional characterization active the components from the medicinal plant, Eugenia jambolana .