There are several sources of information available about the etiology of common complex traits (of which our primary interest is in psychiatric disorders). We have epidemiological, genetic, and evolutionary information to consider in thinking about how genetic and environmental factors interact to influence phenotypes. Epidemiological data tells us something about the prevalence (and its variation within and between populations), its correlation among relatives, and relationships to specific environmental factors. Evolutionary data informs about the rates of mutation, the effects of demographic structure (including assortative mating, migration, drift, inbreeding and the like), and potential effects of natural selection in molding the genetic portion of the etiological architecture of a trait. Genetic data tells us of evidence of linkage (i.e. involvement - or lack thereof - of genes located in certain chromosomal regions), or direct genotypic associations (i.e. etiological effects - or lack thereof - of measured genes, or genes in LD with them) of specific loci. Rarely are these data looked at jointly, however. In this application, we propose to further develop and apply our methods for simultaneously considering all of these data types in an effort to better understand the true etiological architecture. We will work with various simulation-based approaches to consider what phenogenetic models are compatible with the existing data we have about evolution, epidemiology of a given trait, and past attempts at gene finding for those traits (both successful, and, in what we believe is a novel twist, unsuccessful ones as well). Each of these data types informs about what range of etiological models would be plausible for a given disease, although of course there is no way to infer actual truth. Our goal is to eliminate from consideration models, which are inconsistent with existing knowledge, and to compare the power of various study designs and inferential methods under the range of plausible set of etiological models. Models which would have predicted that previous studies should have found the genes would be rejected, as would models inconsistent with our existing epidemiological and evolutionary information. We will explore the set of plausible models and further develop our set of inferential analysis methods for joint linkage and LD analysis in the sort of heterogeneous data structures that we expect will be the most powerful for prospective gene identification in this highly complex psychiatric traits, for which we have massive amounts of data, but unfortunately little real knowledge extracted from them.
The overall aim of this study is to develop and apply new methods to help us better understand elements of the etiology of common diseases such as schizophrenia, bipolar, and Alzheimer disease. These methods, based both on evolutionary and epidemiological inference should allow us to more efficiently exploit the accomplishments of the recent biotechnological revolution for unraveling the complex etiological architecture of these diseases, both through more powerful statistical analysis and optimization of experimental design. It is hoped that better understanding of the true nature of these disorders will be useful for development of prospective public health strategies.
Showing the most recent 10 out of 23 publications
