Visualisation and Analysis of Transcriptional
Networks and Pathways
Tom Freeman
The Roslin Institute, R(D)SVS,
University of Edinburgh
Network Graphs of Biological Relationships
Social interactions between individuals
Transmission of disease
Transcription
Relationship (evolutionary, homology)
between genes and proteins
Interactions between components (pathways,
co-citation, Y2H data, microarray data)
Pathways
Spread of TB via contact tracing
Protein homology
Protein interaction
Microarray Data Analysis
Statistics
Explorative
Graph Paradigm for Gene Expression Data
• Co-expression defined using correlation measure (e.g. Pearson)
• Genes (nodes) are connected to each other in a network based on their level of coexpression (edges)
120
Tissue2
100
95
50
40
4
Gene1
Gene2
Gene3
Gene4
Gene5
Tissue3
50
55
50
50
2
Tissue4
50
60
50
55
5
100
100
100
100
4
gene1
gene2
gene3
gene4
gene5
100
80
Expression
Tissue1
60
40
20
1
3
0
1
2
5
Gene1
Gene2
Gene3
Gene4
Gene5
4
4
Gene1 Gene2 Gene3 Gene4 Gene5
100% 99% 58% 38% 23%
99% 100% 64% 46% 31%
58% 64% 100% 97% 13%
38% 46% 97% 100% 16%
23% 31% 13% 16% 100%
50,000
2
3
Sample
1.25 billion
calculations
50,000
PLoS Comp Biol. 3:2032-42 (2007)
Nature Protocols, 4:1535-50 (2009)
www.biolayout.org
BioLayout Express3D Team
Anton Enright, EBI
Stijn van Dongen, EBI
Thanasis Theocharidis, The Roslin Institute
Tom Freeman, The Roslin Institute
BioLayout Express3D Graph Construction and Analysis Pipeline
Data QC, normalisation and
annotation
Gene to gene Pearson
correlation calculated for every
probe set on the array
Pearson correlations >threshold (0.7)
Filter correlations file based on user
defined threshold (0-1.0)
Edges drawn between nodes (genes) based
on correlations > than selected threshold
Optimised weighted Fruchterman-Rheingold layout
2 or 3D visualisation
MCL clustering, enrichment analysis, exploration
BioLayout Express3D Class Viewer
Ifng time course
a
b
e
f
c
d
Raza et al., BMC Systems Biology. 2:36-51, 2008
GNF Mouse GeneAtlas V3
44 populations of mouse cells
Hume et al., Genomics 2010
R = 0.8, MCL2.2
20,346 nodes, 944,650 edges
Network Modelling of Macrophage Signalling and Effector Pathways
Aims
to construct models of the current consensus view of the biological
pathways underpinning the macrophages role in the innate immune
response
Pathway models should:
• Capture the semantics of relationships between components of the
pathway as described in the literature
• Logically depict molecular interactions using standardised notation
• Easily understood by a biologist
• Be useful in the analysis of genomics data and hypothesis generation
• Provide a resource for the computational modelling of pathways
modified Edinburgh Pathway Notation scheme
Raza et al., BMC Systems Biology 2008
Freeman et al., submitted 2010
http://www.mepn-pathway.org/
IFNG Induction of MHC class2 Antigen Presentation
Drawn using the modified Edinburgh Pathway Notation Scheme (mEPN)
Raza et al., BMC Sys Biol
in press 2010
Dynamic Modelling of Macrophage Pathways
Objectives
• To explore the possibility of
using the pathway diagrams
constructed by us as a resource
for pathway simulation
modelling
• To model pathways activity
using a method that is scalable
and does not require knowledge
of kinetic parameters
Signaling Petri Net:
stochastic flow simulation
• a Signaling Petri Net (SPN) is a Petri net-related method
proposed recently for simulations of biological pathways
See:
Ruths et al. BMC Systems Biology 2:76 (2008)
Ruths et al. PLoS Comp Biol. 4:76 (2008)
• it is an alternate version of Petri net and the algorithm models
the stochastic flow of a variable number of tokens
• it doesn't need kinetic parameters of reactions/transitions
• it’s fast and intuative!
FLOW OF TOKENS THROUGH TRANSITIONS
SPN stochastic flow
SPN Implementation
• Ben Boyer (placement student from Paris) given task of
exploring potential of SPN (Jan, 2009)
• Played with it in PathwayOracle tool, refactored and
implemented SPN algorithm in PIPE2 tool
• Implemented, optimised and parallelised in BioLayout
Express3D code framework (Aug 2009)
• Implemented parsing of graphml model files
• Designed and implemented OpenGL/GPU solution to flow
simulation visualisation (Jan. 2010)
Model in yEd
Model in BioLayout Express3D
Acknowledgements
Pathway Medicine
Paul Lacaze
Peter Ghazal
MSc cohort 2007-2008
www.mepn-pathway.org
The Roslin Institute
Sobia Raza
Mark Barnett
David Hume
QMRI University of Edinburgh
Tamasin Doig
Chris Gregory
BioLayout Express3D
Thanasis Theocharidis, The Roslin Institute
Ben Boyer, The Roslin Institute
Anton Enright, EBI
Stijn van Dongen, EBI
www.biolayout.org
www.macrophages.com
Tom.Freeman@roslin.ed.ac.uk