(A) In this hypothetical example, five extant haplotypes are observed (1–5) within a chromosome segment showing strong LD (low recombination rate). These haplotypes are derived through five mutation steps (resulting in five SNPs in current populations) from the inferred ancestral sequence (boxed in black) and can be grouped into two major haplotype clusters (boxed in green and red). Separating the ancestry-based haplotype clusters are earlier mutation steps (G₃ → A; C₂ → T). Alleles of these SNPs can be applied for “tagging” the clusters (typed in green and red). Currently, ancestry is commonly inferred by either the allele frequencies of SNPs or the corresponding nucleotides in non-human primate species. When the frequency of SNP alleles is applied (preferably using those of African populations), the haplotype clusters are referred to as “haplotype frameworks” [9]. The SNPs tagging the frameworks are termed “framework SNPs” or “fmSNPs”. (B) Tree structure of the five extant haplotypes and the expression phenotype clusters. Given a simple hypothesis that an historical mutation creates a variant altering the expression phenotype (resulting either enhancing or suppressing expression), two alternative schemes of resulting phenotype clusters associated with the variant are illustrated. The left panel exemplifies an evolutionarily earlier expression alteration caused by a mutation tagging the ancestry-based haplotype clusters, and the right panel demonstrates the alteration caused by a more recent mutation (with the mutations boxed and the resulting expression phenotype clusters circled).