Linkage analysis involves statistical inference about the location of
genes influencing a trait, using trait and genetic marker data collected
on families.
We describe a new approach, implemented in a computer program, for
parametric linkage analysis with a quantitative trait model having one
or two genes and a polygenic component, which models additional
familial correlation from other unlinked genes. Competing programs use
simpler models: one gene, one gene plus a polygenic component, or a
crude approximation to the two gene model. Using simple models when
they are incorrect, as for complex traits that are influenced by
multiple genes, can bias estimates and reduce power to detect linkage.
Simulated examples, with various sizes of pedigrees, show that
two-gene analysis correctly identifies the location of both genes,
whereas other analyses based on simpler models fail to identify the
location of genes with modest contributions.
We compute the likelihood with MCMC realization of segregation
indicators at hypothesized gene locations conditional on marker data,
summation over phased multilocus genotypes of founders, and peeling of
the polygenic component. This is the first program for two genes and a
polygenic component. It has no restriction on number of markers or
complexity of pedigrees, facilitating use of more complex models with
general pedigrees.
This is joint work with Elizabeth Thompson and Ellen Wijsman.
Meet the speaker in Room 212 Cockins Hall at 4:30
p.m. Refreshments will be served.