Paper Review: The evolutionary origin of complex features

From: Mathias Ganter (mganter@u.washington.edu)
Date: Tue Nov 30 2004 - 23:05:17 PST

  • Next message: Ravi Kiran: "Paper Review - 1"

    Authors and Title

    Richard E. Lenski, Charles Ofria, Robert T. Pennock, Christoph Adami: The
    evolutionary origin of complex features. Nature 423, 139 - 144 (08 May
    2003); doi:10.1038/nature01568

     
    <http://www.nature.com/cgi-taf/DynaPage.taf?file=/nature/journal/v423/n6936/
    abs/nature01568_fs.html>
    http://www.nature.com/cgi-taf/DynaPage.taf?file=/nature/journal/v423/n6936/a
    bs/nature01568_fs.html

     

    Remarks

    As the title suggests this paper by Lenski et al. simulates the evolution of
    higher organisms from rudimentary ones by the implementation of an
    artificial digital life form that is able to mutate, replicate and compete
    thus following the Darwinian evolution, i.e. natural selection and survival
    of the fittest. The whole digital life is built on computer programs that
    have the ability to perform complex logic functions each of which can be
    used to gain energy.

     

    The main two subjects of this article are mutations and selection of
    superior life forms of digital life.

    Mutations can have various effects (beneficial, neutral, and deleterious).
    Forms of mutations are point mutations, deletions and insertions. They can
    cause damage in the short term, but on the other hand they can ultimately
    become a positive force in the genealogy of a complex organism. The
    complexity results from simplicity, i.e. simple components may yield to more
    extensive ones by the modification existing structures and functions (while
    some are more important than others; in nature, functions can often be
    derived from structure - protein structure predictions to answer functional
    questions). They also mention that the development from simple to more
    complex functions is only possible by using a pre-assigned reward simulating
    Darwinian evolution.

    They also mention that various different mutations can result in various
    solutions for a specific problem, although they represent a different path
    to the goal.

    A great advantage of this digital life implementation compared to real
    evolutionary studies with missing links is the fact that the whole
    evolutionary process including all intermediate states can be analyzed
    easily: replication efficiency and computational merit. There is no lack of
    missing links which makes the analysis easier than using various coalescent
    programs inferring about a whole bunch of unknown parameters like mutation
    rate, branch length, effective population size,.

     

    There is not a real independent evolutionary process because of the fixed
    rewards in form of computational merits.

    In addition, the outcome of the experiment is clearly pre-assigned in form
    of the logical function with the highest reward- in nature, nothing is
    pre-assigned. Are these rewards reasonable or is this prior knowledge biased
    towards the outcome they want.

    Furthermore, the experiment does not represent a higher organism because one
    considers asexual replication, a fixed population size and no recombination
    or migration.

     

    Their studies could be extended by

    - having a population with a varying size,

    - taking several populations into account with migration events,

    - dropping the idea of asexual reproduction and replacing it by sexual
    replication,

    - assigning different rewards,

    - starting with a different set of functions,

    - taking different, more complex logical functions into account.

     

    This problem can be considered as a search problem because one searches for
    the optimal solution in a finite state space (the number of all possible
    genome sequences). It uses a heuristic in form of the rewards.

    It can also be considered as a more or less random walk in a finite state
    space finding the most promising individuals (highest fitness) where each
    individual only depends on its direct ancestor (Markov chain).

     

     


  • Next message: Ravi Kiran: "Paper Review - 1"

    This archive was generated by hypermail 2.1.6 : Tue Nov 30 2004 - 23:05:25 PST