Figure 3

k-means clustering and DBSCAN clustering for Lr34exon11 and BS00083385. k-means clustering and DBSCAN clustering were used to cluster genotypes for each individual on relative read frequency of the two SNP alleles. Genotypes called within the same group are denoted by color. Unfilled symbols indicate samples that were not classified by the algorithms. (A) k-means and (B) DBSCAN clustering of LR34exon11. LR34exon11 locus is a single-copy locus and the two genotypes are easily distinguished by either clustering algorithm. Heterozygotes are characterized by an equal proportion of both alleles. (C) k-means and (D) DBSCAN clustering of BS00083385. This primer pair presumably amplifies multiple loci in the polyploid wheat genome that can still be distinguished based on relative read frequency. The three genotypic classes for individual lines are likely AAAAAA, AABBBB, and AAAABB. The BBBBBB group does not appear to be present, since a null A genotype should fall on the vertical axis with zero reads counts of allele A. DBSCAN did not classify the unfilled individual, which is potentially heterozygous at one of the loci (AAABBB).