Motif Mining from Gene Regulatory Networks

1
Motif Mining from Gene Regulatory Networks

• Based on the publications of Uri Alons group
• presented by Pavlos Pavlidis
• Tartu University, December 2005

2
Gene Regulatory Networks

• From Wikipedia
• Gene regulatory network is a collection of DNA
  segments in a cell which interact with each other
  and with other substances in the cell, thereby
  governing the rates at which genes in the network
  are transcribed into mRNA
• From DOE
• Gene regulatory networks (GRNs) are the on-off
  switches and rheostatsdynamically orchestrate
  the level of expression for each gene.

3
Why networks can regulate Gene Expression?

• U. Alon and his group, stresses the importance of
  the building blocks of the network.
• These building blocks are called motifs

4
Motifs

• They are called also n-node subgraphs in a
  directed graph
• (The work has also been extended for undirected
  graphs)
• They are characterized from the number n of the
  nodes and the relations between them directed
  edges

5
The 13 different 3-node subgraphs
6
Feed Forward Loop
It regulates rapidly the production of Z
7
In what motifs they are interested

• Not in biologically significant
• They dont know a priori if a motif is
  biologically significant
• They can calculate statistical significance
• The probability that a randomized network
  contains the same number or more instances of a
  particular motif must be smaller than P. Here P
  is 0.01.

8
Randomized Network

• A randomized network is not completely
  randomized.It has some properties
• The same number of nodes as in the real network
• For each node the number of the incoming and
  outgoing edges equals to the real network.

9
Representation of the network as a matrix
M Randomization Select randomly two cells which
are 1 e.g A(1,3), B(2,1). If A(1, 1) and B(2,
3) are 0 then swap Goal The randomized network
must have the same sum in columns and in rows
Columns The number of outgoing edges Rows The
number of incoming edges
10
One more requirement If we are looking for
n-node subgraphs, then the number of n-1 node
subgraphs must be the same in real and randomized
networks This is done to avoid assigning high
significance to a structure only because of the
fact that it includes a highly significant
substructure.
11
Significance of a motif

• Three requirements
• P lt 0.01
• P was estimated (or bounded) by using 1000
  randomized networks.
• The number of times it appears in the real
  network with distinct sets of nodes is at least U
  4.
• The number of appearances in the real network is
  significantly larger than in the randomized
  networks Nreal Nrand gt 0.1Nrand (Why??).

12
(No Transcript)
13
What did they find

• That in biological systems as in E.coli or in
  S.cerevisiae only some certain types of motifs
  are statistically important.
• When they studied other systems such asFood
  webs. The database of seven ecosystem food
  websNeuronal networks the neural system of
  C.elegans
• WWW
• OTHER KIND OF MOTIFS WHERE STATISTICALLY
  IMPORTANT

14
FFL SIM DOR
15
FFL

• Biological Example
• the L-arabinose utilization system
• Crp is the general transcription factor and AraC
  the specific transcription factor.

16
The real model
17
FFL

• Coherent
• Incoherent
• Important for the speed of response

18
Software
mDraw              Network visualization
tool (mfinder and network motifs visualization
tool embedded)
19
(No Transcript)
20
(No Transcript)
21
(No Transcript)