|1. Research Institution
|2. Research Area
|3. Research Field
|4. Term of Project
||FY 2000 - FY 2004
|5. Project Number
|6. Title of Project
||Computational Biology on Genome Function Based on Expression and Phenotype Data
7. Project Leader
||Title of Position
||Kyushu University, Faculty of Agriculture
8. Core Members
||Title of Position
||Faculty of Agriculture, Kyushu University
||Kyushu University, Faculty of Mathematics
||The University of Tokyo, The Institute of Medical Science
9. Summary of Research Results
The genomic sequences of many important model organisms are on the verge of being elucidated. Our research goal is to develop technologies critical to post genomic analysis such as transcriptome analysis, proteome analysis and network-based functional analysis to take advantage of this wealth of recently acquired biological information. Furthermore, using information science approaches, we will investigate methods leveraging genome structure to advance from single gene functional analysis to inference of system wide network based functional analysis.
- Expression profile analysis tools
We found a new type of bias, called"print-order-bias". And we propose that the application of an appropriate combination of normalization methods to a microarray dataset not only removes biases effectively but also avoids additional biases to be added onto the expression profile data that may obscure the true expression levels of genes under study. We also introduced functional logistic discriminant analysis which is an extension of the classical method of logistic discriminate analysis to expression profile data.
We have developed multilevel digraph approaches to analysis of Boolean network models. In this method, we first determine according to expression data whether a given gene has influenced the expression of another gene. By systematically altering the alignment, we are able to typify the topology of the vast majority of existing control relationships. Then, by eliminating the effects of indirect gene control influences, we are able to accurately and completely model gene control networks. We also developed methods for graphical network modeling of gene expression control from expression profiling data. Using this methodology, we are able to describe the correlations of genes in the network and then from analysis of weighted associations, we can infer and characterize direct cause-effect relationships among genes.
- Genome annotation supporting system for microbial genome
We developed an annotation supporting database system to greatly increase the speed of processes from sequencing through annotation. Using this system we will able to rapidly annotate the genomic sequence of microbial genomes. To pile up data, a microbial researcher consortium has been organized.
- Expression profile analysis
In our project, we supply enough over 10000 chips for microbial research groups starting post genome research project.
We developed an extensive yeast gene expression library consisting of full-genome cDNA array data for over 500 yeast strains, each with a single gene disruption. Using this data, combined with dose and time course expression experiments with the oral antifungal agent Griseofulvin, we identified CIK1 as an important affected target gene based on Boolean and Bayesian network analysis.
We also show that dedifferentiation in the social amoeba Dictyostelium discoideum relies on a sequence of events that is independent of the original developmental state and involves the coordinated expression of a specific set of genes from expression profile data. These observations establish dedifferentiation as a genetically determined process and suggest the existence of a developmental checkpoint that ensures a return path to the undifferentiated state.
10. Key Words