Implementation of a clustering method for time series gene expression data based on mixed-effects models with Gaussian variables and non-parametric cubic splines estimation. The method can robustly account for the high levels of noise present in typical gene expression time series datasets.