Milestones: April 2005 to April 2009
Statistical Genetic Modelling and Analysis of Complex Traits

Researchers studying complex diseases with a genetic component are increasingly reliant on genetic markers known as single nucleotide polymorphisms (SNPs) for the mapping of susceptibility loci in candidate genomic regions and in genome scans. Our milestones over the next five years focus on the development of novel statistical methods to detect and localize disease genes and to assess candidate genes and environmental factors using SNP data, as well as more conventional microsatellite marker data. 

Develop and evaluate new approaches to identify disease susceptibility genes in nuclear families and in pedigrees using linkage analysis methods that address various forms of heterogeneity and incorporate external information where available.

Develop more flexible models to jointly use allele-transmission and allele-sharing information in family studies designed to localize genes and characterise their effects.

Develop and implement new statistical methods for analysis of associations between complex traits and multi-SNP haplotypes and nongenetic risk factors in the presence of ambiguous linkage phase using data from either cohort or case-control study designs.

Incorporate population-genetic models of the distribution of haplotypes into a maximum-likelihood framework for joint consideration of multiple linked SNPs in the mapping of susceptibility loci in candidate genomic regions.

Develop genetic association models for multiple SNP data that formulate penetrance functions, including gene-environment interactions, in terms of an underlying disease-predisposing locus.

Evaluate the bias and efficiency of environmental effect estimates in family-based case-control studies of candidate genes that use family members as controls.

Apply the methods we develop in on-going national and international studies of complex human traits and diseases.

Complexity of trait expression does not derive solely from multifactorial genetic and environmental etiology, but also from the nature of the phenotype under scrutiny. There is a large class of complex traits for which relevant statistical methodology has just begun to be developed, i.e., function-valued traits, which are infinite-dimensional characters that can be described as a stochastic process (such as antibody profiles after infection, growth trajectories, etc.) with changes governed by a function of certain independent variables, usually non observable, namely, the genotypes. Another class of traits, to some extent overlapping function-valued traits, includes time-to-event phenotypes (such as survival and age-at-onset), recurrent events (such as kidney stone episodes), and interval-censored traits (such as current health status and incubation period when infection time is unknown). Our milestones focus on the development and application of methods for these trait classes.

Develop methodology for estimation and testing that is appropriate for function-valued and time-to-event traits in experimental crosses arising from mouse models of genetic susceptibility to malaria infection (e.g., blood parasitemia profile following infection, time to peak parasitemia, and survival time).

Develop statistical methods for the analysis of studies of longitudinal traits in human populations, in which quantitative and categorical traits are measured repeatedly through time, in order to model dynamics of the disease process, sensitivity to timing of genetic effects, including effect modifier genes, as well as effects of accumulating environmental exposures.

Apply longitudinal trait models to studies of complications of type I diabetes and disease expression in cystic fibrosis.

Refine statistical models and apply to other data sets.

Statistical methods developed for the analysis of quantitative traits in studies of human and animal populations have interesting parallels to those useful in addressing current problems in forest genetics of natural populations, and in understanding questions concerning genetic diversity and forestry conservation genetics. Our milestones focus on the development of statistical models for quantitative trait loci (QTL) for outbred crosses with unknown pedigree.

Develop a computationally intensive (MCMC) sampling algorithm to incorporate pedigree estimation into statistical models for the identification of QTLs associated with tree height data.

Implement methods to select superior families of trees for subsequent breeding programs, based on results of QTL estimation, focussing on the posterior distribution of the maternal genotype at the putative QTL locus and taking into account the planting environment.

Microarray technology, which provides a way to globally measure differential gene expression, chromosomal alterations, and protein levels, promises to be extremely useful for the diagnosis, treatment, and prevention of complex disease, as well as for the elucidation of biological mechanisms. Our milestones focus on the integration of multiple data types in a statistically coherent way to allow for testing of more complex biological hypotheses and richer modelling of biological processes and complex data structures.

Develop measures defining co-regulation of genes, using both mouse recombinant inbred strains and human lymphocyte data.

Develop approaches for combining microarray data and linkage data from recombinant inbred strains of mice.

Develop statistical and computational methods to incorporate individual measures of tissue-specific molecular alterations into genetic studies of cancer families, including comparative genomic hybridization, gene expression, and protein expression microarray data. 

Evaluate the increase in statistical efficiency associated with integration of multiple data types.

©1999 Copyright MSHRI. All rights reserved.