Pairwise comparisons among the haplotypes. A separate line is plotted for each pairwise comparison, but only the different classes of interspecies comparison are assigned different colors. The coordinate system is the same as that in Figure 1. The vertical axis indicates percent divergence between the pairs of haplotype sequences. Pairwise divergences were calculated as averages for 2-kbp windows that slid 500 base pairs between points. The arrows over the gene models indicate the direction of transcription. The most variable exons of the three genes (in all cases, exon 2) are indicated by an asterisk.

Histograms of the pairwise divergence among selected haplotypes. The coordinate system is the same as that employed in Figure 1. Divergences were calculated for 1-kbp bins, based on RefSeq coordinates, counting only discrepancies at nucleotide positions that could be fully aligned with all other haplotypes for which sequence was available in a given region. (A) Representative C-group haplotype compared to a representative E-group haplotype. (B) Recombinant haplotype 04535_1 compared to a representative C-group haplotype. (C) Recombinant haplotype 04535_1 compared to a representative E-group haplotype. (D) A representative B-group haplotype compared to a representative D-group haplotype. The actual haplotypes used to represent haplotype groups B, C, D, and E (see legend to Fig. 1) were 01018_2, 00576_1, 00576_2, and 10923_2, respectively.

Classical DQA1, DQB1, and DRB1 alleles present on the 21 human haplotypes juxtaposed to a molecular phylogeny from the middle of the DQB1-DQA1 intergenic region. The phylogenetic tree is redrawn from Figure 5C with branches leading to haplotype groups A-E, as defined in the Figure 1 legend, labeled. Entries are marked N/A (“not available”) either because there is not a match to an established human allele—as was true for the gorilla and chimpanzee alleles—or insufficient data were available to establish a match.

A visual representation of the multi-aligned haplotypes. The arrows over the gene models indicate the direction of transcription. Haplotype designations are indicated on the left, including membership in haplotype groups A, B, C, D, or E, when appropriate. Haplotypes within a group are nearly identical across the entire region. The coordinate system, in base pairs, is based on haplotype RefSeq (see Methods). Physical distance in the figure corresponds to the longest path through the multi-alignment; deletions relative to the longest path are indicated by gaps in the appropriate haplotypes (and, if space allows, by a series of dots). However, the tick marks and coordinates on the horizontal axis refer to RefSeq coordinates. Coloring of the haplotypes clusters them locally according to the deep branches of the phylogenetic tree to which they belong. The local clusterings were achieved by constructing molecular phylogenies for nonoverlapping 5-kbp windows across the region. Sample phylogenetic trees are shown at the bottom of the figure. The bottoms of the horizontal gray bars across the trees, which correspond to a pairwise-sequence divergence of 4%, indicate the arbitrary threshold for assigning a group of haplotypes the same color in a local region. The gray bars visible at the ends of most haplotypes are segments that could not be assigned a color in that region because they ended within one of the 5-kbp windows used for tree construction. The gorilla sequence, although treated in most instances as a single composite haplotype, is shown here as two overlapping sequences, GORILLA_1 and GORILLA_2, because these two sequences, which are true haplotypes from a single individual, are of different parental origins.

Pairwise linkage disequilibrium between SNPs in the DQB1-DQA1-DRB1 region. The matrix display shows linkage disequilibrium between pairs of SNPs, measured by the statistic |D′| (see Methods). The coordinates correspond to those defined in the Figure 1 legend.

Molecular phylogenies for the 23 haplotypes in four 2-kbp windows sampled from the DQB1-DQA1 interval. The coordinate system is the same as that employed in Figure 1. Parts (A-D) differ only in the position of the 2-kbp window that was analyzed: horizontal bars below the coordinate axes indicate the position of the window that was the source of data for each tree. Branches leading to C- and D-group haplotypes (see Fig. 1 legend) are labeled.