Abstracts from CIS group to be presented at ISMB 2003 http://www.iscb.org/ismb2003/

 

H-52  Improved prediction of protein subcellular localization: exploring inherent biases in neural network learning
Mikael Boden^1
1mikael@itee.uq.edu.au, , School of Information Technology and Electrical Engineering, University of Queensland
Correspondence address: mikael@itee.uq.edu.au

A series of predictors for protein subcellular localization has been based on feed forward neural networks. We propose a set of architectural and algorithmic refinements based on recurrent neural networks and present simulations that improve on previous results.

Long abstract

 

K-14  Towards more biological mutation operators in models of gene regulation
James Watson¹, Nicholas Geard², Janet Wiles
¹jwatson@itee.uq.edu.au, University of Queensland; ²nic@itee.uq.edu.au, University of Queensland
Correspondence address: jwatson@itee.uq.edu.au

Gene regulation is often studied through models of directed graphs. Mutation operators applied to such networks impose limitations on how the models evolve. A method to extract a regulation network from an artificial nucleotide sequence is presented, and the impact of sequence-level mutations on network-level structure is discussed.

Long abstract

 

K-17  Stochastic Neural Network Models for Gene Regulatory Networks
Tianhai Tian¹, Kevin Burrage^2
1tian@maths.uq.edu.au, ACMC, University of Queensland; ²kb@maths.uq.edu.au, ACMC University of Queensland
Correspondence address: tian@maths.uq.edu.au

Stochastic models are presented by introducing stochastic processes into neural network models for studying the genome dynamics. Poisson random variables are used to represent chance events in the processes of synthesis and degradation. Using an example network, we show how to study robustness and stability properties of gene expression patterns.

Long abstract

K-27  Modelling the Role of Small RNAs in Gene Regulation
Nicholas Geard¹, Janet Wiles^2
1nic@itee.uq.edu.au, The University of Queensland; ²j.wiles@itee.uq.edu.au, The University of Queensland
Correspondence address: nic@itee.uq.edu.au

Small functional RNA molecules have been discovered to play an important role in the regulation of gene transcription. The abstract model presented here uses a sequence-matching paradigm to generate regulatory networks that utilise multiple levels of transcriptional control to increase their computational power.

Long abstract

 

K-28  High level properties of genetic regulatory network
Kai Willadsen¹, Janet Wiles^2
1kaiw@itee.uq.edu.au, University of Queensland; ², University of Queensland
Correspondence address: kaiw@itee.uq.edu.au

Abstract models of gene regulation date back to the development of the Random Boolean Network model in 1969. This class of models aims to investigate emergent properties of genetic regulatory networks with a view to better understanding high-level characteristics of the behaviours that these systems display.

Long abstract

 

K-46  Controlling Complexity in Biological Networks
Bradley Tonkes¹, Janet Wiles ², John S. Mattick
¹btonkes@itee.uq.edu.au, Institute for Molecular Bioscience, School of Information Technology and Electrical Engineering, University of Queensland; ²janetw@itee.uq.edu.au, School of Psychology, School of Information Technology and Electrical Engineering, University of Queensland
Correspondence address: j.mattick@imb.uq.edu.au

A major challenge for modeling and representing biology in silico is to map real biological processes onto computational space. In this paper we introduce a new approach based on (artificial) recurrent neural networks, which allows the modeling of control systems and dynamical trajectories of differentiation and development.

Long abstract