Twenty brewing characteristics which contribute to classify a
Kyokai sake yeast strains were selected by the partition analysis for 24
Kyokai yeast strains which contained 385 stock strains.
It was possible to classify these as independent clusters for
Kyokai yeast strains using 20 explanatory variables. This narrowed to 344 strains when excluding 41 strains of samples missed from the 24 main clusters.
While the total numbers of distinction error strains of a discriminant analyses using 18 explanatory variables for 344 strains were 7 strains, when considered with 16 explanatory variables which excluded the amino acid component left 10 strains. Therefore, we decided to exclude amino acid composition from the analysis items because the contribution was small and there was little influence in cluster analysis and discriminant analysis.
Regardless of the old or new period distributions, the brewing characteristics of the 344 strains of K-7 were kept for only one cluster and the k-7 cluster could be roughly divided into 3 sub clusters by the cluster analysis. The reason for being divided into 3 clusters was because of the fermentation ability.
In the cluster analyses by 11 brewing characteristics for 24 kinds of all yeast selection candidates in 1250 strains since 2003, 76 strains of clusters of high aroma generation strains were sorted out and the composition strains were only K-1601, K-1701 and K-1801. Also, all K-1801 distribution strains from 2006 to 2014 were contained in the cluster of K-1801, and the brewing characteristics of the K-1801 were stable.
View full abstract