EPSRC Network on Evolvability in Biology & Software Systems

Evolvability, Genetics & Development in Natural and Constructed Systems: Abstracts of the EPSRC Evolvability Network Symposium

Tewin Bury Farm Hotel, Hertfordshire, England, UK
26-28 August 2003

University of Hertfordshire Computer Science Technical Report 389
C. L. Nehaniv, P. J. Bentley & S. Kumar (Editors)

Artificial EvoDevo as a Proposed Tool for the Validation of Genomic Research

Artificial EvoDevo as a Proposed Tool for the Validation of Genomic Research


Adaptive Systems Research Group
University of Hertfordshire
Hatfield Herts AL10 9AB
United Kingdom


Fifty years ago a small publication appeared in Nature about the structure of the DNA molecule, which, according to one of the authors might be of considerable interest for the understanding of the genetic mechanism. Fifty years later, in the same journal, a very much more impressing looking review article was published about the considerable consequences of that seemingly innocuous paper. It heralds the definite start of the Genomic Era and the reader is treated to the stupendous insights and knowledge molecular genetics has brought and the promises it has for improving human well-being. However, one of the main conclusions of this and related papers is that the genome is a much more complicated system than anticipated and that we do not have the tools and methods to adequately process the massive flow of data that modern technology provide us with. To study the workings of the genome we collect data and use statistical/computational procedures to make sense of them. The problem is that the inferences drawn from the applications of these methods depend on underlying assumptions about the genomic mechanism, which was the subject of investigation in the first place. Some of these assumptions are about the constancy of the evolutionary rate (to reconstruct phylogenetic trees), the homogeneity of the structure of the DNA (with respect to regions of mutability), the tacit denial of epigenetic mechanisms and the linearity of many population genetics models (to make them mathematically more tractable). Especially evo-devo research, with its emphasisis on dynamics, has shown that many of these assumptions simply do not hold. Actually, the fact that they don't might have given us more insight in the workings of the genome than sequencing in its own right.

A worrying aspect, however, is that because of the demonstrable failure of the assumptions, the conclusions drawn from traditional genomic science might be misleading. This is the more serious given the conglomerate of business, technology and governmental policies in which genomics is embedded and the social and political implications it therefore might have.

For that reason I will advocate a research program that focuses especially on the effect of the assumptions on the validity of the tools we use in genomics. The idea is to create populations of reproducing and evolving artificial creatures of which the phenotypic expression is realized by virtual genome dynamics (in concert with a simple metabolism). The model should be set up in such a way that it allows for systematic substitution of the traditional assumptions by newer findings from evo-devo research and for tracking the consequences of such operations. To enable the latter, the creatures are to be analyzed as if they were ``real" living organisms by means of currently employed methods, i.e. by sequencing their DNA, using algorithms to detect hot spots of binding sites, constructing phylogenetic trees to elucidate their evolution and by applying the statistics of quantitative genetics (for example after breeding experiments). Apart of being a benchmark to check the correctness of the methods, these artificial creatures are also tools to develop new hypotheses.