Variable predictions of protein quality by chemical score due to amino acid analysis and reference pattern.

The amino acid composition of five protein samples (casein, beef, wheat flour, peanut flour and soy protein isolate) was obtained from several sources: various routine analytical procedures, manufacturer's fact sheets, and published data. The amino acid profile from each source was scored against three reference patterns, each based on an estimate of human essential amino acid requirements. Protein quality ratings based on chemical score were variable. Both the chemical score and predicted first-limiting amino acid for a given protein often differed with data source and choice of reference pattern. Some of the predictions contradicted those previously validated by bioassays. The variability in protein quality ratings was reduced somewhat by expressing the data as essential amino acid indices. But this also reduced the degree of difference in protein quality among the various proteins. Although the amino acid profile of food proteins provides an important perspective on their nutritive value, it appears that the imprecision and uncertain accuracy of routine hydrolysis and analytical procedures confound the use of amino acid data for regulating protein quality. The situation is further complicated because human essential amino acid requirements have been variably interpreted, which has led to different reference patterns.