The genetic algorithms in computer science is based on the old
Mendelian genetics. I'd like to know more about the abstract model of
DNA/genetics/evolution based on our current knowledge about genes:

Here is a good example of an abstract model:
- Typogenetics (by Douglas Hofstadter in Godel, Escher, Bach)
http://www.csse.monash.edu.au/hons/projects/1999/Andrew.Snare/

However, this model is too basic and it doesn't model important
behaviors like gene regulation and phenotype development. But I like
this example because it describes the high-level logic about DNA,
instead of going into the molecular details.

---

Request for Question Clarification by hedgie-ga on 03 Jul 2002 08:32 PDT

There are two different computer  applications related to genetics:

1) Genetic algorithm or GA which is an algoritm for finding minimum
ofa  complex function
  - inspired by medelian genetics -  but not attempting to model real
genes

2) Computer model of DNA and  inheritance. Today several models of
different aspects
    and processes exist.

Which of the two are you interested in/

---

Clarification of Question by kenlaw-ga on 08 Jul 2002 11:19 PDT

I'm interested in Genetic Algorithm. However, I'd like to see how we
can improve it by borrowing the insights from the latest genetics, and
possibly from the computer models of DNA and inheritance. I hope this
will clarify my question.

Thanks for  a great question,  Kenlaw

  In his comment rmg provided some interesting links. 

  The GA field is huge, as apparent from directories such as 
 http://dmoz.org/Computers/Artificial_Intelligence/Genetic_Programming/Algorithms/
or 
http://www.aic.nrl.navy.mil/galist/

 and so the challenge was to select  few  papers which mention
enhancement inspired by biology.

  My search strategy was to combine 'Genetical Algorithm' ,
'evolutionary computing', ..
 with terms from biological genetics: Cross-over, protein folding,
chromosome ,...

I found few references  and am quoting  the abstracts and follow by
URL pointing to the article:

 Development, Learning and Evolution in Animats - Kodjabachian, Meyer
(1994)
 The work of Boers and Kuiper [BOER92] combines a genetic algorithm
 and a learning procedure with a
 of Boers and Kuiper [BOER92] combines a genetic algorithm and
 a learning procedure with a Lindenmayer
 designer or by an automatic process inspired from biology
 and involving the three main adaptive processes
 www.biologie.ens.fr/fr/animatlab/perso/kodjaba/jkjamperac.ps.gz



Grammatical Evolution: A Steady State approach. - Conor Ryan (1998)
  Michael.ONeill@ul.ie ABSTRACT We describe a Genetic Algorithm that
can evolve co
mplete programs.
ABSTRACT We describe a Genetic Algorithm that can evolve complete
programs. Using a
the chromosome. In a manner similar to natural biology, 
shine.csis.ul.ie/papers/fea98.ps

:The project described in this paper attempts to apply insight gained
from studying biological systems to a conventional Genetic Algorithm.
Natural genes are contained on a chromosome, surrounded by long
sequences of "waste" DNA. Useful genes are selected fro
m amid
the waste by the operation of RNA polymerase. Natural selection also
occurs within relatively large groups; small populations tend
 to die out from a lack of genetic variation. We believe both
of these features are relevant in the... (Update)
A Few New Features For Genetic Algorithms - Marshall Graves  
(Correct)
biological systems to a conventional Genetic Algorithm. Natural genes
are contained on a chromosome,
www.vuse.vanderbilt.edu/~gravesm/papers/GAFeatures/GAFeatures.ps

Adapting Crossover in a Genetic Algorithm (1995)

Abstract: Traditionally, genetic algorithms have relied upon 1 and
2-point crossover operators. Many recent empirical studies, how
ever, have
shown the benefits of higher numbers of crossover points. Some of the
most intriguing recent work has focused on uniform crossover
, which
involves on the average L/2 crossover points for strings of length L.
Despite theoretical analysis, however, it appears difficult
to predict when a particular crossover form will be optimal for
a given problem.
http://citeseer.nj.nec.com/spears95adapting.html



 The search strategy, and use of citiacion index such as
http://citeseer.nj.nec.com
 is likely to produce more papers of this type. I did not try to be
exhausitve.

 Good way to keep up with the developement in what can be called a
subfiled of GA
would be to follow journals such as

 Advances in Complex Systems 
http://www.tbi.univie.ac.at/ACS/Vol2.html
 
   See e.g. paper in Vol2, issue 1
RNA In Silico: 
... with keywords  RNA Secondary Structures, Fitness Landscapes,
Energy Landscapes, Molecular Evolution, Punctuated Equilibrium,
Folding Kinetics, Folding Pathways
http://www.tbi.univie.ac.at/ACS/Abstracts/ACS_V2_I1_3.html

or  (expensive) Journal of theoretical biology 
http://www.academicpress.com/jtb

   I hope these link answer your question, and please feel free to ask
for
   more specific clarification. It was an interesting assigment.

Hedgie

Thanks for looking into it! I've read the Boers' paper mentioned in
your answer, and it is very useful. I'll spend some time looking at
your answer. If you find other interesting stuff, I hope you can send
me an update through this page.

http://ds9a.nl/amazing-dna contains a lot of relevant links that may
help you. It also explains parts of DNA from a computer perspective.

Thanks for the link. It seems to be interesting. I'll look at it in details.

The following page at George Mason University has a list of
publications by members of their Genetic Algorithms group. All the
publications are downloadable, and a few  seem to involve a more
sophisticated biological approach. (Two possibly promising titles are
"The Coevolution of Antibodies for Concept Learning," and "The Role of
Mutation and Recombination in Evolutionary Algorithms.")

http://cs.gmu.edu/research/gag/pubs.html

The following articles may be worth your while, but I couldn't find
them online.

Cangelosi, A. and J. Elman (1995). Gene regulation and biological
development in neural networks: An exploratory model. Technical Report
CRL-UCSD, University of California, San Diego.

Narayan Behera and Vinyanand Nanjundiah. Trans gene regulation in
adaptive evolution: A genetic algorithm model. Journal of Theoretical
Biology, 188(2):153-162, 1997.

If you can find a copy of Stuart Kaufmann's book _The Origins of
Order_, I think there is a discussion of neo-Darwinism and genetic
algorithms in Chapter 1. (Look at it before you buy it, because I'm
not sure.)

Finally, here is a paper called "Contextual Genetic Algorithms." The
abstract sounds like it might involve both the gene regulation and
phenotypic development you mentioned: "A genetic algorithm scheme with
a stochastic genotype/phenotype relation is proposed. The mechanisms
responsible for this intermediate level of uncertainty, are inspired
by the biological system of RNA editing found in a variety of
organisms. In biological systems, RNA editing represents a significant
and potentially regulatory step in gene expression. The artificial
algorithm here presented, will propose the evolution of such
regulatory steps as an aid to the modeling of differentiated
development of artificial organisms according to environmental,
contextual, constraints. This mechanism of genetic string editing will
then be utilized in the definition of a genetic algorithm scheme, with
good scaling and evolutionary properties, in which phenotypes are
represented by mathematical structures based on fuzzy set and evidence
theories." The paper is at

http://www.c3.lanl.gov/~rocha/e95_abs.html

I hope at least one of these leads proves fruitful. Good luck!