Biospectroscopy and partial least squares regression as a sensitive tool for optimisation of milk quality

Fourier transform infrared (FTIR) technology enables cheap and fast characterisation of milk samples. This is currently used for monitoring of parameters such as total fat and protein content in cattle milk. We believe that much more information can be drawn from the spectra, for instance about fatty acid composition. The recent availability of TINE's FTIR-database at Campus Ås, in combination with a very good cattle recording system in Norway, enables detailed studies of pedigree and environmental influences on a large scale. Partial least squares regression (PLSR) is used to predict fatty acid composition in milk from a designed feeding experiment. Many informative structures are revealed. The calibration model and similar models will be used in studies of the Norwegian cattle population. The poster presents some explorative results in a study which will include pedigree information as well as genomic, transcriptomic, or metabolomic data. Trained sensory panels may also provide quality parameters important in consumer studies.