8 - Transcription Factor CREM Dependent Expression During Mouse Spermatogenesis
Tim Beissbarth, Igor Borissevitch, Andreas Hoerlein, Annette Klewe-Nebenius, Bernhard Korn, Martin Vingron, Guenther Schuetz, German Cancer Research Center
Transcription factor CREM seems to be the trigger of expression at late stages of spermatogenesis. Large-scale cloning, sequencing, and expression profiling of messages expressed in CREM dependent manner was performed: 1) Subtractive cloning (SSH) of messages absent in CREM-/- mice; 2) Sequencing the obtained library and clone selection; and 3) Gene expression profiling.

9 - Comparing the Similarity of Time-series Gene Expression Using Signal Processing Metrics
Atul J. Butte, Children’s Hospital, Boston; Ling Bao, Massachusetts Institute of Technology; Ben Y. Reis, Timothy W. Watkins, Isaac S. Kohane, Children’s Hospital
Treating gene expressions as discrete time-invariant signals, we "tuned" the matrix of 2,467 yeast genes at 18 time points revealing gene-gene associations with particular phase shift and gain. We found 18 associations (two known in the literature). All are ranked poorly using conventional clustering. Signal processing can enhance clustering algorithms.

10 - Classification of Cancer Tissue Types by Support Vector Machines Using Microarray Gene Expression Data
Jinsong Cai, Columbia University; Aynur Dayanik, Rutgers University; Hong Yu, Naveed Hasan, Tachio Terauchi, William Noble Grundy, Columbia University
A support vector machines classifier was used to classify cancer and normal tissues based on DNA microarray gene expression patterns with ~99% accuracy. A list of genes (100 of the total 4,026 genes) whose expression profiles had the best correlations with tissue types was identified using the Fisher discriminant criterion.

11 - 2HAPI, A Comprehensive, Internet-Based Microarray Data Analysis System
J. Lynn Fink, Michael Gribskov, University of California, San Diego
2HAPI is a system for computational microarray data analysis that attempts to create an integrated analytical environment that is highly accessible, fully-featured, and free to academic users. 2HAPI is designed with the notion that the user need not be a computer scientist or statistician.

Poster 12 moved .

13 - Unfolding Expression Data from cDNA Microarrays
Andrew Goryachev, Princess Margaret Hospital, Canada; Pascale Macgregor, Aled Edwards, University of Toronto
We present a mathematical model describing relationship of measured fluorescent intensities and actual mRNA concentrations. Employing methods of robust statistics we developed algorithms allowing one to estimate parameters of the model and apply unfolding transformation extracting actual ratios from raw data. We also discuss methods for measurement of ratio noise and reproducibility of experiments.

14 - Comparison of Statistical Tests in the Analysis of Differential Expression in Hybridization-based Experiments
Ralf Herwig, Pia Aanstad, Matthew Clark, Hans Lehrach, Max-Planck Institut für Molekulare Genetik
We compare statistical tests in their performance on detecting differential expression on cDNA arrays. Evaluation of experimental and simulated data indicates that the analysis via repeated hybridization experiments followed by statistical testing is an accurate and sensitive way to identify even small expression changes (1:1.5) on a large scale.

15 - A Practical Filtering Procedure of Gene Expression Data for Clustering Analysis
Christian Hödar, Verónica Cambiazo, INTA University of Chile; Chris Vulpe, University of California, Berkeley; Mauricio González, INTA University of Chile
We measured and filtered the expression values of each gene in a mouse cDNA microarray to diminish their initial number, and to compare their relative expression levels between two experimental conditions. We found 86 genes that changed in one condition, 80% of them remained associated within five main clusters.

16 - General Toolkit of the E-cell System for Modeling Gene Expression System
Kenta Hashimoto, Sae Seno, Fumihiko Miyoshi, Masaru Tomita, Keio Universsity
We present a generic model of gene expression and regulation for the E-CELL system, a general purpose cell simulator. Using this generic model, we have simulated: (1) regulation of the lac operon in E. coli, and (2) the lytic-lysogenic switch network in bacteriophage lambda.

17 - Efficient Data Processing Method for Large-scale cDNA Microarray Analysis
Koji Kadota, Yasushi Okazaki, Hidemasa Bono, Rika Miki, Kentaro Shimizu, Yoshihide Hayashizaki, RIKEN Tsukuba Institute, Japan
It is important to obtain a reproducible data set when analyzing the expression profile using cDNA microarray. We have developed an efficient data processing method from duplicated experimental results. We applied this method to the tissue expression profiling data and will present the feasibility and importance of our method.

18 - A Comparative Analysis of Gene Expression Profiles in Yeast
Özlen Konu (Grantham), Ming D. Li, University of Tennessee
Comparative studies are essential for determining the degree to which gene activity is compartmentalized in response a particular treatment. Using publicly available yeast microarray data, we identified different sets of genes that were solely responsive to either heat- or cold-shock. Functional profiles of these heat- and cold-shock specific gene sets also were compared.

19 - Expression Profiling on cDNA Arrays: A Robust Method for Resolving Hybridisation Intensities into Background and Positives
Hilmar Lapp, Marion Weissmann, Gudrun Werner, Novartis Research Institute Vienna
For the analysis of cDNA array hybridisations the intensity values are assumed to reflect the abundance level of a transcript in the sample, which is only valid for signals significantly different from noise. We devised a method that robustly resolves the population of observed intensities into two populations, namely background and positives.

20 -Probe Design on Genomic Level for High-density DNA Oligo Microarray
Fugen Li, Gary D. Stormo, Washington University School of Medicine
High-density DNA oligo microarrays are widely used in biomedical research. In this poster we describe algorithms to optimize the selection of specific probes for each gene in an entire genome for oligo chips. We test the algorithm with a few model organisms.

21 - Comparative Analysis of Yeast Microarray Gene Expression Data: Hierarchical Clustering and Self-organizing Maps
Jinfeng Liu, Lei Shi, William Grundy, Columbia University
We analyzed a set of yeast gene expression data by hierarchical clustering and self-organizing maps (SOMs). According to the MIPS Yeast Genome Database (MYGD), clusters were obtained by minimizing a defined cost function. Only a few MYGD classes could be clustered by either method. Hierarchical clustering performed slightly better than SOMs when evaluated both externally and internally.

22 - Definition of Hidden cDNA Array Gene Expression Profile Tracks—Application to Molecular Diagnosis of Ovarian Cancer
Ann-Marie Martoglio, James W. Miskin, David J. C. MacKay, Stephen K. Smith, University of Cambridge
We present a novel approach to array-based gene expression data that allows for "blind" separation of samples based on hidden gene expression profile tracks (gTRACKS). The method is demonstrated on data from tailored cDNA array ovarian cancer studies and shows successful application for various classifications of the tissue samples.

23 - Fundamental Patterns Underlying Gene Expression Profiles: Simplicity from Complexity
Smita Mitra, Neal Holter, Amos Maritan, Marek Chiplek, Jayanth Banavar, Nina Fedorff, Pennsylvania State University
Singular value decomposition analysis of previously published microarray expression data has uncovered underlying patterns in their expression profiles. The essential features of the profiles are captured using just a small number of these patterns, leading to the striking conclusion that the transcriptional response of a genome is orchestrated in a few fundamental patterns of gene expression change.

24 - Comparison of Normalisation Methods for Microarray Expression Data Over Multiple Experiments
Norman Morrison, Magnus Rattray, University of Manchester; Kenneth Pollock, Ray Jupp, Aventis, UK; Andy Bras, University of Manchester
We have compared a number of different normalisation methods for expression data derived from different technological systems. Raw data was normalised by a combination of approaches using a range of noise cut-off points. The approaches were assessed by a metric describing the strength of biological signal post-normalisation.

25 - Studying Network Dynamics in Genetic and Neural Systems
Brendan Mumey, Tomas Gedeon, Julie Taubman, Zuzana Gedeon, Kate Hall, Montana State University
Learning regulatory elements in genetic networks based on gene expression data collected from DNA microarrays is exciting; Bayesian networks are one proposed model. We examine a different network model based on a more restricted network trajectory graph and demonstrate our methods using gene expression data sets and neural activity trace data that we suggest as another testbed for network-learning algorithms.

26 - Combining Microarray Expression Data and Phylogenetic Profiles to Learn Gene Functional Categories Using Support Vector Machines
Paul Pavlidis, William Noble Grundy, Columbia University
We demonstrate how to apply the support vector machine learning algorithm to a heterogenous data set. Our results suggest that combining data types should only be attempted if there is evidence that the functional classification of interest is clearly reflected in both data sets.

27 - Clustering Analysis of Gene Expression Data Using Percolation
R.Sasik, T. Hwa, University of California, San Diego
We present a novel method for clustering of gene expression data based on the percolation paradigm. In this method the result is cast in terms of the probability that a gene belongs to a certain cluster, accommodating for the possibility that it participates in several clusters or none at all.

Poster 28 withdrawn by author.

29 - Supervised Learning of Microarray Expression Profiles: Analysis of an Acute Leukemia Data Set as an Example
Patrick D. Sutphin, Soumya Raychaudhuri, Russ B. Altman, Stanford University
We used a publicly available supervised learning tool developed in our lab to analyze leukemia data. Clustering methodologies failed to distinguish between the subtypes of leukemia; acute myeloid and acute lymphoid leukemia do not cluster separately. Our supervised classification strategy was effective in differentiating between the two subtypes.

30 - Expression Profiler: An Integrated Tool for Gene Expression and Sequence Analysis
Jaak Vilo, Alvis Brazma, Alan Robinson, Wellcome Trust Genome Campus, UK
Expression Profiler is a collection of Web-based tools aimed for analysis of gene expression data from DNA microarray projects. Its four components perform clustering and visualization of expression data (EPCLUST), extraction of genome-specific information and sequences (GENOMES), submission of clustering results to other tools (URLMAP), and sequence pattern discovery (SPEXS). http://ep.ebi.ac.uk/

31 - A Fuzzy Logic Approach to Analyzing Gene Expression Data
Peter J. Woolf, Parke-Davis Pharmaceutical Research; Yixin Wang, University of Michigan, Ann Arbor
We developed a fuzzy logic algorithm to transform gene expression values into qualitative descriptors that is evaluated by using heuristic rules. A model to find triplets of activators, repressors, and targets in a yeast data set was tested. This extends techniques such as clustering in that it generates connected networks of genes.

32 - Validating Clustering for Gene Expression Data
Ka Yee Yeung, David R. Haynor, Walter L. Ruzzo, University of Washington
Many clustering algorithms have been proposed to analyze gene expression data. We provide a systematic and quantitative framework to assess clustering results. Our methodology is to apply a clustering algorithm to the data from all but one experimental condition, and to use the remaining condition to assess clustering results.

33 - Discovery: A Tool for Class Prediction Using Gene Expression Data
Louxin Zhang, Zhuo Zhang, Song Zhu, Kent Ridge Digital Labs and Bioinformatics Centre, Singapore
The poster introduces a Web tool for analysis gene expression data and class prediction, which is roughly like diagnosis: given a set of known classes, determine the correct class for a new sample. The method has great potential in improving cancer classification and diagnosis by using gene expression data.