PaGE - Patters from Gene Expression


Patterns from Gene Expression

PaGE 5.1 Released September 2nd, 2004
Note: A Java version of PaGE is available now. Please check it out.

What is PaGE?

PaGE is free downloadable software for microarray analisys. PaGE can be used to produce sets of differentially expressed genes with confidence measures attached. These lists are generated the False Discovery Rate method of controlling the false positives.

But PaGE is more than a differential expression analysis tool. PaGE is a tool to attach descriptive , dependable, and easily interpretable expression patterns to genes across multiple conditions, each represented by a set of replicated array experiments.

The input consists of (replicated) intensities from a collection of array experiments from two or more conditions (or from a collection of direct comparisons on 2-channel arrays). The output consists of patterns, one for each row identifier in the data file.

One condition is used as a reference to which the other types are compared. The length of a pattern equals the number of non-reference sample types. The symbols in the patterns are integers, where positive integers represent up-regulation as compared to the reference sample type and negative integers represent down-regulation.

The patterns are based on the false discovery rates for each position in the pattern, so that the number of positive and negative symbols that appear in each position of the pattern is as descriptive as the data variability allows.

The patterns generated are easily interpretable in that integers are used to represent different levels of up- or down-regulation as compared to the reference sample type.

To illustrate this,  the following table gives an excerpt of data for four of the gene tags in a given of hybridization experiment and four sample types.  There are three replicates for sample types G0 and G2 and two replicates for sample types G1 and G3. As they are these data are hard to peruse for information.
 
gene tag
G0 I
G0 II
G0 III
G1 I
G1 II
G2 I
G2 II
G2 III
G3 I
G3 II
1
0.0114
0.0328
0.0151
0.0060
0.0236
0.0436
0.5640
0.8920
0.0639
0.2490
2
0.0050
0.0131
0.0061
0.0041
0.0364
0.0296
0.8830
0.7000
0.0199
0.1050
3
0.0629
0.2340
0.0431
0.2270
0.2120
0.0105
0.1400
0.0243
0.0117
0.0907
4
0.0250
0.0600
0.0264
0.1500
0.2660
0.0134
0.1860
0.0851
0.0172
0.0112
 
If G0 is used as a reference sample type, the patterns attached by PaGE to these tags might look like

gene tag
G1
G2
G3
1
0
7
2
2
0
8
1
3
2
-1
-1
4
3
1
0

this is an easily interpretable set of patterns. For example gene tag 3 is detected as up-regulated 2 levels in sample type G1 and down-regulated one level in sample types G2 and G3, as compared to sample type G0.

Publications, Errata Corrige and Notes

* Grant G.R., Liu J., Stoeckert C.J.Jr. (2005) A practical false discovery rate approach to identifying patterns of differential expression in microarray data, Bioinformatics, Vol 21 no 11, 2684-2690.

Note: There is a typo in this publication on page 2686 in formula for muk(i+1) (the second to last displayed forumla on the page). On the right hand side, the first muk(i) should be mu-tildek(1) (i replaced by 1).

* Grant G.R., Liu J., Stoeckert C.J.Jr. The technical manual for PaGE 5.1.

* Grant G.R., Manduchi E., Stoeckert C.J. Jr. Using non-parametric methods in the context of multiple testing to identify differentially expressed genes. Methods of microarray data analysis, editors S.M. Lin and K.F. Johnson, Kluwer Academic Publishers (Boston, 2002): 37-55. (Winner of the best presentation award CAMDA'00).

* Manduchi E., Grant G.R., McKenzie S.E., Overton G.C., Surrey S., Stoeckert C.J. Jr. (2000) Generation of patterns from gene expression data by assigning confidence to differentially expressed genesBioinformatics, 16(8): 685-698.

  Errata Corrige and Notes to the above original paper
 

Software

To download the perl version of the PaGE 5.1 software click here. This is the latest stable release of PaGE and presents several improvements as compared to the previous release (4.0). Besides bug fixes, it offers a more informative and richer output and at the same time it is easier to use as it has less mandatory options.

To download the Java executable for the PaGE 5.0 software click here. This is the first java implementation of PaGE. It should be considered a Beta version and will be updated shortly to version 5.1 which will match all of the features of the perl version 5.1.

If you have questions on the program or its usage or if you want to report any bugs, please contact: ggrant@pcbi.upenn.edu.



 
 
 
PCBI Homepage Last updated: September 30, 2004
If you have questions or comments contact: ggrant@pcbi.upenn.edu
CBIL Homepage