Using multivariate techniques to predict wheat flour dough and noodle characteristics from size-exclusion HPLC and RVA data

Flour proteins of hard and soft winter wheats grown in Oregon were characterized by size-exclusion HPLC (SE-HPLC). Flour pasting characteristics were assessed by a Rapid Visco Analyser (RVA). Principle component scores (PCS) were calculated from both RVA data and from absorbance area and % absorbance values from SE-HPLC. The PCS and cross-products, ratios, and squares were used to derive wheat classification and quality prediction models. A classification model calculated from PCS of SE-HPLC data could reliably separate these hard and soft wheats. The prediction models for mixing and noodle characteristics showed better performance when calculated from PCS values of both SE-HPLC and RVA data than from SE-HPLC data only. The R2 values of prediction models for mixograph absorption, peak time, and tolerance were 0.827, 0.813, and 0.851, respectively. Prediction models for noodle hardness, cohesiveness, chewiness, and resilience immediately after cooking had R2 values of 0.928, 0.928, 0.896, and 0.855, respectively. These results suggest that multivariate methods could be used to develop reliable prediction models for dough mixing and noodle characteristics using just SE-HPLC and RVA data.