Tree scanning: a tool for studying the genetic architecture of
complex traits
Alan Templeton
Washington University
Tree scanning is a method to use evolutionary trees of haplotypes to
study phenotypic associations. It is applicable to genomic regions
with little to no internal recombination. The haplotype diversity in
such regions can be described by a haplotype tree. Tree scanning
exhaustively examines all possible bi-allelic partitions of the
tree. If the first scan detects significant associations, additional
rounds of tree scanning are used to partition the tree into three or
more allelic classes. I first review published work on an analysis
of associations between variation in various lipid traits and genetic
variation at the Apolipoprotein E (APOE) gene in three human
populations. Tree scanning successfully identified phenotypic
associations expected from previous analyses. Tree scanning for the
most part detected more associations and provided a better biological
interpretative framework than single SNP analyses. Tree scanning
detected genetically determined phenotypic heterogeneity within all
three electrophoretic allelic classes. Next, I show how tree
scanning can be extended to analyze multivariate traits, longitudinal
data, and epistasis between any two genomic regions separated by much
recombination. Overall, tree scanning is a simple, powerful and
flexible method for using haplotype trees to detect
phenotype/genotype associations at candidate loci and to define
genetic architecture in terms of epistasis and pleiotropy among
candidate loci.