Rapid cultivar identification of barley seeds through disjoint principal component modeling

Iain Whitehead, Alicia Munoz, Thomas Becker

Research output: Contribution to journalArticlepeer-review

Abstract

Classification of barley varieties is a crucial part of the control and assessment of barley seeds especially for the malting and brewing industry. The correct classification of barley is essential in that a majority of decisions made regarding process specifications, economic considerations, and the type of product produced with the cereal are made based on the barley variety itself. This fact combined with the need to promptly assess the cereal as it is delivered to a malt house or production facility creates the need for a technique to quickly identify a barley variety based on a sample. This work explores the feasibility of differentiating between barley varieties based on the protein spectrum of barley seeds. In order to produce a rapid analysis of the protein composition of the barley seeds, lab-on-a-chip micro fluid technology is used to analyze the protein composition. Classification of the barley variety is then made using disjoint principle component models. This work included 19 different barley varieties. The varieties consisted of both winter and summer barley types. In this work, it is demonstrated that this system can identify the most likely barley variety with an accuracy of 95.9% based on cross validation and can screen summer barley with an accuracy of 95.2% and a false positive rate of 0.0% based on cross validation. This demonstrates the feasibility of the method to provide a rapid and relatively inexpensive method to verify the heritage of barley seeds.

Original languageEnglish
Pages (from-to)773-783
Number of pages11
JournalAnalytical and Bioanalytical Chemistry
Volume409
Issue number3
DOIs
StatePublished - Jan 2017

Keywords

  • Barley classification
  • DPCM
  • LoaC capillary electrophoresis
  • PCA

Fingerprint

Dive into the research topics of 'Rapid cultivar identification of barley seeds through disjoint principal component modeling'. Together they form a unique fingerprint.

Cite this