Table 1

Transformation of data for regression analysis

RESPONSE VARIABLE
 PREDICTOR VARIABLE
A. Absolute deviation from a reference state
 yi =
xi = Xi - Xir
B. Relative deviation from a reference state

C. Lotka-Volterra system
xi = Xi

We assume the general linear model is y = ai0 + Σ(aij xj). The Xi denote experimental time series data for metabolite i, while the slopes () are estimated from the smooth output functions of the artificial neural network that had been trained on the experimental data. Subscript r denotes the value of the metabolite at a reference state. Linearization options I and II are included in transformations A and B respectively, assuming that the reference state is a steady state. For a piecewise linear linearization (option III), the data may be transformed following either A or B.

Veflingstad et al. Theoretical Biology and Medical Modelling 2004 1:8   doi:10.1186/1742-4682-1-8