Exploiting most informative markers to predict group membership of North Aegean sheep

Aim of the present study was to discover a small panel of SNPs with the highest discriminatory power to classify sheep to: (i) three neighboring North Aegean islands (Lemnos, Lesvos and Agios Efstratios) and (ii) two agro-ecological zones (rough and flat landscape) within Lemnos. A total number of 256 ewes belonging to n=15 herds dispersed on the three islands and the two agro-ecological zones were genotyped with the 50K SNP array. Application of quality criteria at the marker level resulted in about 37K SNPs retained. Of these markers, we searched for a limited number (n=50 or 100) of strongly differentiated SNPs between groups (islands/zones) defined as those with highest values for the F ST fixation index and combined results of principal components (PCA) and Nonparametric Discriminant Analysis (NPDA) to derive the classification criterion (DC) across cases. Application of the DC derived on the first two principal components (PCs) constructed on n=100 strongly differentiated SNPs resulted in a lowest average misclassification error rates (0.019) across the islands, whereas the respective DC derived on the first PC of n=50 strongly differentiated SNPs could accurately assign individuals to the two agro-ecological zones (average misclassification error rate: 0.012). Current results suggest that the proposed approach could be implemented to unequivocally determine the origin/habitat of local sheep and possibly their associated specific products.