Model functional properties of proteins from amino acid composition.

It has long been of interest to model the functional properties of proteins, e.g., their contribution to foaming, lipid binding, gel formation and the like, as a function of their physico-chemical properties--hydrophobicity, surface activity and viscosity. In some cases, it has been possible to model protein and peptide properties based on their content of relevant amino acids and those acids' principal properties. These had been based on the principal component analysis of amino acid chemical and physico-chemical property data. The principal properties were expressed as three z-scores.

Because peptide physico-chemical properties stem from their content of specific classes of amino acids--acidic, basic and nonpolar amino acids--it should be possible to model the functional properties of proteins directly from amino acid composition. Scientists at Cornell University attempted to model these protein functional properties based on their complete amino acid composition and amino acid principal properties: three term z-scores or five term zz-scores.

The scientists modeled previously published measurements of protein physicochemical and functional properties from the sums of the principal property contributions of each amino acid to each principal property. They used the Partial Least Squares Regression method to do this. For each of the principal properties, the moles of each amino acid in the protein were multiplied by the corresponding principal property score for that amino acid, and then summed.

Models based on the contribution of each of the 20 coded amino acids to each of the five zz-score sums were developed to predict protein hydrophobicity (R=0.940), viscosity (R=0.738) and foam capacity (R=0.942). This research shows that protein functional and physicochemical properties can be successfully estimated from amino acid composition using five, not three, principal properties. Knowing this information may help us determine the foaming properties as well as other functional properties of proteins.

Further information. Karl Siebert, Department of Food Science and Technology, Cornell University, New York State Agricultural Experiment Station, 129 Food Research Laboratory, Geneva, NY 14456; phone: [phone]; fax: [phone]; email: [e-mail].