Russell Standish wrote:
I'm trying a slightly different tack with Tierra, of artificially
inducing mass extinctions every now and then.
Cool. A limitation of even `advanced' genetic phylogenetic inference
techniques is that they are not only ignorant of protein structure (what
DNA changes are likely to be viable), and ignorant of recombination, but
they are typically even ignorant of the notion of time as a directional
concept. There is no population in these models so there can be no
fierce or relaxed environment for the population.. What these models
can estimate, they do so by assuming the randomness of many simple
independent mutational processes (e.g. a mutational process for each
available and comparable amino acid in the organisms of interest), to
estimate the ordering and time between bifurcation events. If there is
some strong selection pressure acting on part of a protein, the model of
diversification will be biased by that. If there are several strong
selection pressures it will be more biased.
But you can implement recombination, you can capture directional
selection, and track genotype/phenotype mapping. It seems clear to me
that simulations are a good way to think about these issues in the
abstract (i.e. life as it could be). But it's not obvious to me
if/how they can say much more than phylogenetic techniques for
understanding the real world of contemporary DNA, unless there are other
kinds of historical data to constrain the simulations... ?
Marcus
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
lectures, archives, unsubscribe, maps at http://www.friam.org
