Starch molecular structure and phosphorylation investigated by a combined chromatographic and chemometric approach

Structural features of starch were studied with special emphasis on the rel ationship between starch phosphorylation and starch chain length distributi on comparing a chemometric approach with classic statistics. Starches prepa red from 44 plant species were analysed with respect to the degree of phosp horylation and chain length distribution of the neutral unit chains, prepar ed by enzymic isoamylase debranching, using high performance anion exchange chromatography with pulsed amperometric detection (HPAEC/PAD). Chemometric algorithms such as Principle Component Analysis (PCA), Non-Negative Altern ating Least Squares Regression (NNALSR) and Partial Least Squares regressio n (PLS) were used to analyse the systematic variation in the chromatograms and compared to Gauss decomposition. Detailed relations between chain lengt h data and structural elements of the current cluster/bocklet model of nati ve starch granules were revealed. By using PCA, both crystal polymorphs and botanical origin of the starch were accurately predicted. Using PLS, a str ong correlation (0.93) was obtained between the chain length distribution a nd the degree of phosphorylation in the potato starch group. The use of che mometrics as an efficient tool to classify and predict starch functionality is documented. (C) 2000 Elsevier Science Ltd. All rights reserved.