[BioPython] Abstract for BOSC

[Brad Chapman]

Hello all!

Thanks to Andrew's reminders and the big black marking on my Monty
Python calendar that says "Send in your BOSC abstract, dummy!" I
managed to get together a preliminary abstract for BOSC. I'm planning
on talking about some code that I wrote over this past semester for
genetic algorithm and neural network stuff (and which I also hope to
get together for possible inclusion in Biopython is you all like
it!).

Attached is the preliminary abstract for this -- if anyone has time to 
look it over and send comments, etc that would be very appreciated.

Additionally, I'd like to second Andrew's encouragement to submit
abstracts for BOSC -- it'd really be great to see a big Biopython/python
presence at the conference, and to learn what everyone is working on.

Brad


Genetic algorithm and neural network libraries for biological analysis 
in python

Brad Chapman

Department of Crop and Soil Sciences; University of Georgia

Computational Intelligence techniques such as Neural Networks and
Genetic Algorithms have become popular for solving complex problems
without making assumptions about the underlying system being
analyzed. Many biological problems fall into the category of problems
that can be analyzed using these techniques, thanks in part to the recent
explosion of data generated by genome sequencing projects. In an
attempt to use computational intelligence techniques for biological
problems of interest, a set of libaries was developed in the python
programming language. In addition to the backbone code necessary for 
analysis, supplemental code was also developed to allow rapid
conversion of biological sequence data into a form that can be
directly used in a genetic algorithm or neural network. The design and 
implementation of these libraries will be discussed, with examples of
their usage. Additionally, the use of this library for whole genome
analysis will be presented, using the example of finding Matrix
Attachment Regions in the Arabidopsis genome. Discussion will center
around the types of problems which are readily approachable using 
neural networks or genetic algorithms.