Differential Expression Effector Prediction (DEEP)

Category Genomics>Gene Expression Analysis/Profiling/Tools

Abstract Differential Expression Effector Prediction (DEEP) is a tool that takes your data on differential ‘gene expression’ (i.e. SAGE or microarray data) and predicts additional molecules which may be of importance in either of the two tissues (or conditions or time points...) under examination.

It does this by combining your data with information on signaling networks stored in the TRANSPATH® database (see G6G Abstract Number 20121).

DEEP consists of a core server module and a JSP-based web interface, both running at the manufacturer's site.

Additionally, you can download and install a Java client for DEEP on your local computer.

Since the local client allows for much more interactivity when inspecting the calculation results (e.g. zooming and panning), its use is strongly encouraged.

DEEP Background and Motivation --

A wide range of technologies, like Serial Analysis of Gene Expression (SAGE) or microarrays, are available for obtaining data on the expression levels of genes in different tissues and/or under different conditions.

Though yielding valuable information about the involvement of genes in processes specific for the different systems under examination, most of the data analysis methods available to-date neglects the fact that a gene doesn't necessarily have to be differentially expressed in order to exert effects, specific for e.g. a certain tissue or disease state.

Rather, its gene product may just be subject to functional modulation by other, differentially expressed genes, thus functioning as the actual effector of a specific expression profile.

The manufacturer presents a method (and provides an implementation thereof, called DEEP - Differential Expression Effector Prediction) to identify such molecules by combining gene expression data with biological expert knowledge about biomolecular interaction networks provided by resources like the TRANSPATH database on signal transduction (Biobase GmbH, Wolfenbüttel).

DEEP Calculation and Output --

In principle, DEEP is capable of processing both SAGE data (i.e. lists of gene frequencies) and gene lists with corresponding P values, as provided by microarray experiments.

As a first step, the user may either select two subsets of publicly available SAGE libraries (e.g. various libraries from normal vs. carcinoma tissues) or upload their own SAGE data.

Each subset is merged into a meta library prior to the identification of genes which are ‘differentially expressed’ according to a user-defined significance threshold.

In the next step, the identified genes are mapped to signaling molecules as described by TRANSPATH ortholog molecule entries.

A signal transduction network is then reconstructed by stepwise expansion up to a user-defined search depth, starting from the identified TRANSPATH molecules, yielding a graph with molecules as nodes and signal transduction events as directed edges.

In a second graph traversal procedure, the starting nodes' significance levels for differential expression percolate to their -- Not necessarily differentially expressed! -- Successors.

If a successor node has multiple incoming edges, the corresponding starting nodes' significance levels are summed up after weighting them by a factor that decreases with increasing distance.

Consequently, each molecule node's value is based on the significance by which one or more genes were considered differentially expressed in a certain experiment, yielding a measure for the degree to which its activity may be influenced in a tissue- or dignity-specific manner.

Finally, the graph is visualized, representing each molecule node's (initial or calculated) ‘significance value mapped’ to a color spectrum ranging from red (specific for tissue/condition 1) over yellow (Not tissue/condition-specific) to green (specific for tissue/condition 2).

Thus a node's, or sub-network's, predominant color immediately gives a ‘visual clue’ on which molecules - differentially expressed or Not - may play pivotal roles in the tissues or conditions under examination.

These nodes, identified in this manner, could therefore represent e.g. potential target molecules for therapeutic intervention and rational drug design.

System Requirements

DEEP has been implemented in the ‘Java programming language’ and consists of three (3) parts: the DEEP core server, a JSP-based web interface, and a downloadable Java client.

This architecture prevents users from having to take care of maintenance tasks, like updating the utilized databases.

Just like the browser-based solution, the Java client communicates with the core server via the firewall-friendly Hypertext Transfer Protocol (HTTP).


Manufacturer Web Site DEEP

Price Contact manufacturer.

G6G Abstract Number 20510

G6G Manufacturer Number 104128