Krasnow Institute > Monday Seminars > Abstracts

Based on the known history of human and animal pandemic viruses, we hypothesized that there is a direct and necessary relationship between the evolvability of a virus and its potential for emergence and epidemic spread. To examine this hypothesis we constructed a computational model of biological viral replication called "VIV (Virtual Virus)" based on a genetic algorithm. In a typical VIV simulation, a population of bit strings evolves for thousands of generations through iterative rounds of replication, mutation, recombination, and selection for fitness. Guided by the molecular biology and molecular epidemiology of the Human Immunodeficiency Virus (HIV), we built features of biological viral structure and replication into the basic genetic algorithm, including (1) a diploid genome, (2) "many to one" genotype to phenotype mapping (3) multiple reading frames (4) inter-strand recombination (5) sequence homology-driven cross-over and (6) variable length. So as to simplify interpretation of results, phenotypes were expressed as strings of the interpretation of results, phenotypes were expressed as strings of the English letters A to Z with start and stop signals, and fitness was measured as correct spelling of the words "envelope", "polymerase", and "coreprotein" anywhere along the genome. Results were plotted as learning curves of aggregate spelling score (for the env, pol, and core genes) against generation number.This simple prototype model exhibited several unanticipated yet biologically plausible properties:

• Recombination in any form accelerates adaptation
• homologous recombination is superior to random recombination
• Fitness improves initially by genome lengthening with multiple gene copies, then by genome shortening to remove non-contributory genes
• Each adaptation problem has its own optimal mutation rate which is very close to 1 mutation per genome.
• At mutation rates above one per genome, genome lengths vary directly with the

mutation rate, with frequent multiple gene copies.
The relevance of this model to known epidemic virus diseases will be discussed.

Back to Top

The Krasnow Institute for Advanced Study
Mail Stop 2A1, George Mason University, Fairfax, VA 22030
Phone: (703) 993-4333 Fax: (703) 993-4325
Email: krasnow-webmaster@gmu.edu