Application of Structural Equation Modelling for Oil Accumulation System Control in Oleaginous yeast

Abstract

Recently, we developed a new statistical method for revealing the regulatory systems in living cells. Our method is based on Structural Equation Modelling combined with factor analysis. In generally, Structural Equation Modelling is utilized to detect the model adaptability with the measured data, such as large-scale questionnaire data. In this study, we improved our developed iteration algorithm and gene selection procedure to infer the causalities between variables as a regulatory network from limited numerical data. Our improved gene selection method is based on cross correlation to summarize the time preceding information from gene expression profiles, which were systematically measured at 8 time points. Cross correlation is usually utilized as a measure of similarity between two waves by a time-lag application, and we defined the values of lags ranging from -2 to +4. By this improved method, we selected 14 genes as regulatory factors for the specific system in oleaginous yeast. In the inferred model, only 6 genes among the selected 14 genes were considered to affect the volume of oil accumulation in a closed and specific system. Our method will be useful to artificially control cell systems in the bioproduction and biotechnology fields.