A procedure for the clustering of cell wall mutants in the model plant Arabidopsis based on Fourier-transform infrared (FT-IR) spectroscopy

FT-IR microspectroscopy can be used to study the global composition and architecture of plant cell walls and it allows cell wall mutants to be identified. Our aim is to define a distance between cell wall mutants in the model species Arabidopsis based on FT-IR spectra. Since the number of data points that constitute a spectrum exceeds the number of samples analysed, it is essential to reduce first the dimension of the dataset. We present a comparison of several compression methods, including linear discriminant analysis using a non-canonical covariance matrix. The calculated distances were used to define clusters of mutants that appeared to be biologically meaningful.