Colloquium Abstract

Thursday, February 24, 2005 • 3:30 PM • Robert N. Noyce Conference Room

Jon Wilkins Bauer Center for Genomics Research, Harvard University

Constructing realistic models using information from different disciplines: case studies in genomic imprinting and human genetics

In trying to understand biological systems, it is necessary to build realistic models that capture the most important and interesting features of those systems. However, in doing this, it is easy to lose any predictive or discriminatory power by complicating models to the point where they can be fit to any empirical pattern. The risk of overfitting therefore sets an upper limit on the degree of realism that can be incorporated into models. Fortunately, this limit can often be extended by incorporating information from other disciplines.

Accurate and detailed modeling can best be accomplished by taking a problem-centered approach, as opposed to a technique-centered one. I will discuss two examples from my own work: (1) evolutionary models of intragenomic conflict that use information from molecular and developmental biology, and (2) a new project on non-equilibrium processes in human demographic history, where more realistic population genetic models can be built using information from archaeology, anthropology, and geology.

Genomic imprinting: intragenomic conflict and coevolution

Many evolutionary processes are driven by conflict, such as competition for mates or food. Evolutionary conflicts also arises within individual organisms when natural selection acts differently on different sets of genes. One such conflict occurs in mammals and flowering plants between the genes that came from the mother, and those that came from the father (the maternally and paternally derived alleles). Due to the possibility of multiple paternity (either within litters or over the course of the mother’s lifetime), paternally derived alleles are selected to demand more resources from the mother than maternally derived alleles are. Simple game-theoretic analyses of this conflict have been successful in explaining certain aspects of the evolution of “imprinted genes,” where the expression behavior of an allele depends on its parent of origin. For example, imprinted genes that increase the demand for maternal resources are expressed only from the paternally derived allele; genes that suppress demand are expressed only maternally.

However, certain features of genomic imprinting are not understandable in terms of a two-player game involving maternally and paternally derived alleles. The establishment, propagation, and interpretation of genetic imprints involves interactions among at least six sets of genes, each of which is evolving in response to distinct selective forces. Determination of the payoff functions for each of these genetic factions is not sufficient to determine the outcome of this multi-player evolutionary game. It is only possible to make meaningful predictions by using information from molecular and developmental biology to define the set of strategies available to each faction. The model then accounts for specific features in the mechanisms by which imprinting is established and regulated.

Non-equilibrium population dynamics in human genetics

A similar problem arises in analyses of human genetic diversity, which are typically based on simple models of population structure. Many of these analyses make various equilibrium assumptions that apply poorly to human history. For example, ethnographic data suggest historical changes in demographic patterns that can only be accounted for with non-equilibrium models. Specifically, different modes of subsistence (foraging, pastoralism, etc.) and political organization (tribal versus state societies) are associated with different patterns of marital residence (matrilocality versus patrilocality) and reproduction (rates of polygyny, etc.). Technological and political innovations will therefore have altered patterns of migration and reproduction in a spatially and temporally heterogeneous manner. The genetic consequences of these heterogeneities will be evident in contemporary human diversity.

In this case, it may be possible to constrain more sophisticated models of human history by incorporating information from archaeology, anthropology, and geology into the genetic analyses. As a simple example, I will discuss some recent genetic results on sex-biases in human migration. Different studies of worldwide patterns of mitochondrial and Y-chromosome genetic diversity have generated substantially different conclusions. The patterns of marital residence in the ethnographic record suggest a non-equilibrium model of migration that may be able to explain this discrepancy in terms of the sampling schemes used in the different studies.