Skip Navigation


Bioinformatics Advance Access originally published online on December 1, 2005
Bioinformatics 2006 22(3):367-368; doi:10.1093/bioinformatics/bti778
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (Print PDF) Freely available
Right arrowOA All Versions of this Article:
22/3/367    most recent
bti778v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Google Scholar
Right arrow Articles by Leung, E.
Right arrow Articles by Bushel, P. R.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Leung, E.
Right arrow Articles by Bushel, P. R.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Published by Oxford University Press 2005.
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press are attributed as the original place of publication with the correct citation details given; if an article is subsequently reproduced or disseminated not in its entirety but only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact journals.permissions@oxfordjournals.org

PAGE: phase-shifted analysis of gene expression

Elo Leung 1 and Pierre R. Bushel 2,*

1Bioinformatics and Computational Biology, School of Computational Sciences, George Mason University Manassas, VA 20110, USA
2National Institute of Environmental Health Sciences, National Center for Toxicogenomics, Microarray Group, Research Triangle Park NC 27709, USA

*To whom correspondence should be addressed.


   ABSTRACT
TOP
 ABSTRACT
1 INTRODUCTION
 2 METHODS AND IMPLEMENTATION
 3 APPLICATION FEATURES
 4 RESULTS FROM APPLICATION
 5 REQUIREMENTS FOR PAGE
 REFERENCES
 

Summary: Grouping of gene expression patterns across biological experiments, treatments and time-series data is performed in q-intervals of measurements using phase-shifted analysis of gene expression (PAGE); a Java-based tool to find clusters of genes that share trends of expression profiles within the dataset. The patterns and genes within q-Clusters are visualized in trend plots and compared to determine biological relevance from the gene annotations.

Availability: PAGE is available at http://dir.niehs.nih.gov/microarray/software/page/

Contact: bushel{at}niehs.nih.gov

Supplementary information: The Supplementary data are available at http://dir.niehs.nih.gov/microarray/software/page/


   1 INTRODUCTION
TOP
 ABSTRACT
 1 INTRODUCTION
2 METHODS AND IMPLEMENTATION
 3 APPLICATION FEATURES
 4 RESULTS FROM APPLICATION
 5 REQUIREMENTS FOR PAGE
 REFERENCES
 
Many methods that identify the correlation between gene clusters are designed to handle series data, which is derived from a biological condition and time or dose treatment. These methods include Cross-Correlation Function, Needleman–Wunsch alignment algorithm, and q-Clustering (Ji and Tan, 2005). Cross-Correlation Function and Needleman–Wunsch use scoring functions that are based on every pair of genes over all time points in the dataset. Both methods are computationally expensive and cannot locate the time interval at which the gene has high correlation with another gene. The q-Clustering method overcomes these limitations by using q-consecutive intervals to determine the correlation across and within the genes. An algorithm called rhythmic analysis of gene expression (RAGE) was developed to extract and characterize gene expression profiles from genome-wide microarray data based on periodic biological processes (Langmead et al., 2003). Similar to the former two methods, the current q-Clustering implementation and RAGE are limited to gene expression data from a dose or time-series experiment. Many datasets such as those generated for toxicogenomics, pharmacogenomics or clinical\disease\genetic informatics involve gene expression measurements across biological type, dose and time (Hamadeh et al., 2004; Hubner et al., 2005). A method that computes the correlations between gene clusters should be able to analyze data from biological condition, treatments and time points simultaneously so that the patterns can be compared.

A Java-based software, phase-shifted analysis of gene expression (PAGE), was developed to analyze gene expression data across multiple biological conditions, treatments and time series. PAGE was applied to a trans-compound dataset to extract the phase-shifts of gene expression in response to seven different toxicants at various dose treatments and time points.


   2 METHODS AND IMPLEMENTATION
TOP
 ABSTRACT
 1 INTRODUCTION
 2 METHODS AND IMPLEMENTATION
3 APPLICATION FEATURES
 4 RESULTS FROM APPLICATION
 5 REQUIREMENTS FOR PAGE
 REFERENCES
 
The gene expression data for analysis with PAGE were generated using rat liver samples analyzed on the Agilent RatTox oligo array. The data are available for download at http://dir.niehs.nih.gov/microarray/datasets/home-pub.htm and will be submitted to the Chemical Effects in Biological Systems Database (http://cebs.niehs.nih.gov/) as a subset of the NCT Compendium dataset. Java 2 Standard Edition and JFreeChart were used for development of PAGE.

The 433 differentially expressed gene patterns were first selected on a per agent, dose and time point manner, ranked as informative using relevance analysis and then filtered for genes with profiles that contain missing values.

The PAGE method is based on the q-Clustering method designed to identify time-lagged patterns of genes in time-series gene expression data (see Ji and Tan, 2005 for details). Briefly, the PAGE method has the following three phases:

  • Phase 1: Gene expression pattern matrix transformation into –1,0,1 to indicate the direction of expression change from each biological condition at fixed time points and treatments. All biological replicates are averaged if provided.
  • Phase 2: Generate q-clusters that have similar patterns of expression of over q-consecutive conditions.
  • Phase 3: Assign a significance score for each bicluster in all q-Clusters and identify the inhibition patterns of each q-Cluster.

A window size of 3 and default threshold of 1.0 was used for specifying the q-interval consecutive points and the bin cut-off for defining the upward and downward trends, respectively. As the normalization threshold approaches zero, the bin for the unchanged gene expression from one point to the next shrinks, in turn the upward and downward gene expression bins expand, and vice versa.


   3 APPLICATION FEATURES
TOP
 ABSTRACT
 1 INTRODUCTION
 2 METHODS AND IMPLEMENTATION
 3 APPLICATION FEATURES
4 RESULTS FROM APPLICATION
 5 REQUIREMENTS FOR PAGE
 REFERENCES
 
PAGE uses a line graph to dynamically illustrate the phase-shifted patterns of gene expressions based on the q-Cluster selected. Each line shown on the line graph represents the trend of a bicluster whose score is equal to or below the maximum threshold value. The line graph can be zoomed in and can be exported in jpg format. Also, the genes associated with the trends shown on the line graph can be exported to a text file. Furthermore, all phase-shifted patterns in each of the q-Clusters can be exported as a tab-delimited text file.


   4 RESULTS FROM APPLICATION
TOP
 ABSTRACT
 1 INTRODUCTION
 2 METHODS AND IMPLEMENTATION
 3 APPLICATION FEATURES
 4 RESULTS FROM APPLICATION
5 REQUIREMENTS FOR PAGE
 REFERENCES
 
4.1 q-Clustering at the gene level
The q-Clusters generated from the differentially expressed genes contain expression profiles that have the same and opposite patterns of expression. The average of the gene expression profiles in each bicluster is shown in Figure 1 to illustrate the trend similarity across four of the seven toxicants, which were determined to have patterns in phase-shift. Figure 1a shows the upregulated trends of gene expression profiles from acetaminophen, allyl alcohol, carbon tetrachloride and methapyrilene treatments, whereas Figure 1b shows the downregulated trends from these treatments. As shown in Figure 1b, the downregulated trends across the patterns of acetaminophen, allyl alcohol, carbon tetrachloride and methapyrilene suggest similarity in biological response at the gene level. These results are in agreement with Waring et al. (2001) but they also reveal patterns of genes from the methapyrilene time-series experiment, which are phase-shifted with the patterns of genes from the acetaminophen dose–response experiment.


Figure 1
View larger version (31K):
[in this window]
[in a new window]
  		Fig. 1 PAGE user interface and visualization of q-clusters. (a) is a screenshot of the PAGE user interface that contains panels for listing projects and q-Clusters, filtering and visualizing the bicluster scores, displaying the phase-shifted patterns within the q-Clusters (q-Cluster ID: 4 shown), and selecting the trends to be visualized or for subsetting the genes from the selected trends. (b) is a plot of the patterns within q-Cluster ID:8. The y-axis is log base 2 intensity and the x-axis is the index of the experiment treatment points. Shown are q-Clusters of window size 3 and the phase-shifted patterns of all the trends are indexed from 0 to 2.

 
4.2 Pathway mapping
The union of the genes from the two q-Clusters in Figure 1 was obtained for pathway mapping using DAVID (Dennis et al., 2003). Results indicate that folate biosynthesis is upregulated and fatty acid metabolism is downregulated. These results suggest the potential response mechanism(s) since the genes in these q-Clusters correlate with the acetaminophen biological response as revealed previously by Heinloth et al. (2004).


   5 REQUIREMENTS FOR PAGE
TOP
 ABSTRACT
 1 INTRODUCTION
 2 METHODS AND IMPLEMENTATION
 3 APPLICATION FEATURES
 4 RESULTS FROM APPLICATION
 5 REQUIREMENTS FOR PAGE
REFERENCES
 
The gene expression data should be normalized and transformed to the user's satisfaction. In addition, missing data need to be imputed or profiles containing them removed, and all treatments and time points need to be consecutive with at least one interval (two adjacent points). The software has been tested on Windows PCs running the XP operating system and requires JRE version 1.4.2 or later.


   Acknowledgments
 
The authors thank Maribel Bruno and Frank G. Bottone, Jr for review of this manuscript and testing of the software. This research was supported by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences. Requests for software should be addressed to E.L.

Conflict of Interest: none declared.


   FOOTNOTES
 
Associate Editor: Steen Knudsen

Received on August 30, 2005; revised on October 27, 2005; accepted on November 11, 2005

   REFERENCES
TOP
 ABSTRACT
 1 INTRODUCTION
 2 METHODS AND IMPLEMENTATION
 3 APPLICATION FEATURES
 4 RESULTS FROM APPLICATION
 5 REQUIREMENTS FOR PAGE
 REFERENCES
 

    Dennis, G., Jr, et al. (2003) DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol, . 4, P3[CrossRef][Medline].

    Hamadeh, H.K., et al. (2004) Integration of clinical and gene expression endpoints to explore furan-mediated hepatotoxicity. Mutat. Res, . 549, 169–183[ISI][Medline].

    Heinloth, A.N., et al. (2004) Gene expression profiling of rat livers reveals indicators of potential adverse effects. Toxicol. Sci, . 80, 193–202[Abstract/Free Full Text].

    Hubner, N., et al. (2005) Integrated transcriptional profiling and linkage analysis for identification of genes underlying disease. Nat. Genet, . 37, 243–253[CrossRef][ISI][Medline].

    Ji, L. and Tan, K.L. (2005) Identifying time-lagged gene clusters using gene expression data. Bioinformatics, 21, 509–516[Abstract/Free Full Text].

    Langmead, C.J., et al. (2003) Phase-independent rhythmic analysis of genome-wide expression patterns. J. Comput. Biol, . 10, 521–536[Medline].

    Waring, J.F., et al. (2001) Clustering of hepatotoxins based on mechanism of toxicity using gene expression profiles. Toxicol. Appl. Pharmacol, . 175, 28–42[CrossRef][ISI][Medline].


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (Print PDF) Freely available
Right arrowOA All Versions of this Article:
22/3/367    most recent
bti778v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Google Scholar
Right arrow Articles by Leung, E.
Right arrow Articles by Bushel, P. R.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Leung, E.
Right arrow Articles by Bushel, P. R.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?