Software Tool: GENIUS

GENIUS: a new tool for gene networks visualization
Paolo Ciccarese, Stefano Mazzocchi, Fulvia Ferrazzi, Lucia Sacchi

Methods for gene network reconstruction based on : (Reverse engineering methods)

• Boolean networks
• Bayesian networks
• Differential Equations

INPUT:
For n genes in the network, an nXn matrix such that
aij = 1 if connection between genes i and j
aij = 0 if no connection between genes i and j

GENIUS visualizes
Genes = nodes
connections = edges

Two types of visualizatiobs:

AGORA STYLE
Algorithm used assumes that every individual can be treated exactly the same.
Simulation paradigm: "PRIVATE SPACE"
This mathematical model uses a repulsive force field and a basic attractive force field.
- Repulsive force field --> Infinity
as
distace between objects --> 0
- Then
Repulsive force field rapidly decreases to 0 on a short distance.
- Attractive force field starts with 0 and increases to infinity.

The Agora view tool has been extended so that a connection between two genes is directed such that the 'from node' is the regulator and the 'to node' is the regulated gene.

TOUCHGRAPH STYLE
This view is useful to show relationships between nodes characterized by maximum level of the number of edges in the minimum-length path connecting these nodes in the graph.

This paper then examines the network visualization of cDNA microarray data set analyzed in [1] and then analyzing temporal profiles relative to the 517 genes using the Reveal algorithm described in [2].Data set available here.

Brief description of Reveal algorithm
- For every gene x,
find set of regulators(minimal set of input genes that can univocally explain behavior of output gene x)
- Based on use of Entropy and Mutual Information scores:
if for 2 genes x and y,
Mutual Information(x,y) = Entropy(x)
then
y univocally determines x

Use of Reveal in GENIUS
They extend the algorithm to include 3 discretization data levels instead of 2.
-1 : under-expression
0 : equal expression
+1 : over-expression
of serum stimulated cell genes w.r.t. expression values of same genes measured using non-stimulated cells.

179 groups(pseudo-genes) recognized and extended algorithms was applied to them.

You can check out the example given in the paper to see how the output looks.

REFERENCES
1. Iyer V. R. et al. (1999): The transcriptional program in the response of human fibroblasts to serum. Science: 283: 83-87
2. Liang S, Fuhrman S, Somogyi R. REVEAL, a general reverse engineering algorithm for inference of genetic network architectures. Pacific Symp. Biocomp. 1998: 98 (3):18-29.