Alleles clustered according to nucleotide similarity.

<p>(A) Phylogenetic tree reconstruction of the gene segments in the haplotype sequenced in [<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1005117#pcbi.1005117.ref013" target="_blank">13</a>] (which is the GRCh37 haplotype). Circles highlight alleles that are evolutionarily very close. Tree made using neighbor-joining method in ape package in R based on Hamming distance between multiple sequence alignment. (Phylogenetic reconstruction using BEAST [<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1005117#pcbi.1005117.ref035" target="_blank">35</a>] led to a qualitatively similar tree). Allele numbers are suppressed for clarity. (B) Distribution of percent nucleotide difference (Hamming distance divided by alignment length) between alleles from same IMGT segment (blue) compared against alleles from different segments (green). Alleles from duplicate segments (e.g. 1-69 and 1-69D) have been merged for this analysis. (C) Same as (B) but with alleles partitioned by operationally defined gene clusters rather than IMGT segment name. (D-F) Heatmaps of matrices of Hamming distance between alleles. Rows and columns are ordered according to gene clusters found by hierarchical clustering as described in Materials and Methods. Color spectrum ranges linearly from red to white for nucleotide distances 0-10%. Differences greater than 10% are white. (D) Alleles from family 1. (E) Alleles from family 3. Full set of alleles in <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1005117#pcbi.1005117.s003" target="_blank">S3 Fig</a>. (F) Alleles in family 4. Dashed white squares indicate possible gene clusters.