1
Standardized visualization of differences between model versions
Tom Gebhardt, Martin Scharm, Vasundra Touré, Dagmar Waltemath, Olaf WolkenhauerDepartment of Systems Biology and Bioinformatics, University of Rostock
COMBINE 2016 Newcastle upon Tyne
23rd of September 2016
m nthe
computational modeling in biology network
2
Motivation: models are evolving
Biological models are constantly evolving implying that several versions of a model can exist.
Scharm et al. (2015) An algorithm to detect and communicate the differences in computational models describing biological systems - Bioinformatics.
Figure courtesy: Tom Gebhart
3
Motivation: models are evolving
Biological models are constantly evolving implying that several versions of a model can exist.
Statistics based on two repositories of computational models: Physiome and Biomodels.
Number of models: 3237
Number of model versions: 14439
Number of deltas: 10662
Scharm et al. (2015) An algorithm to detect and communicate the differences in computational models describing biological systems - Bioinformatics.
http://most.sems.uni-rostock.de/
1
3
7
20
55
148
403
1097
2981
8103
22026
59874
162755
Cha
nges
Figure courtesy: Tom Gebhart
4
Motivation: visualizing models
Plus, it is common to have a visual representation of biological networks…
Systems Biology Tools for Integrated Omics Analysis - M. Hugues
H. Kim (2015), Nature Chemical Biology
Brown et al. (1999), Trends in Microbiology
5
Motivation: visualizing models
Plus, it is common to have a visual representation of biological networks…
… but understanding them depends on the reader’s knowledge …
Ras recruits
RafRas
translocates Raf
Ras activates
Raf
Systems Biology Tools for Integrated Omics Analysis - M. Hugues
H. Kim (2015), Nature Chemical Biology
Brown et al. (1999), Trends in Microbiology
Ras Raf
6
The Systems Biology Graphical Notation
… unless the Systems Biology Graphical Notation is used.
SBGN webpage: http://sbgn.github.io/sbgn/
standardized glyphs and arcs and semantics specific to each.
sinksource
SBGN webpage: http://sbgn.github.io/sbgn/
7
The Systems Biology Graphical Notation
… unless the Systems Biology Graphical Notation is used.
SBGN webpage: http://sbgn.github.io/sbgn/
standardized glyphs and arcs and semantics specific to each.
sinksource
Process DescriptionBiochemistry
Activity FlowMolecular biology
Entity RelationshipPhysiology, Genetics
Three languages are defined.
SBGN webpage: http://sbgn.github.io/sbgn/
8
The Systems Biology Graphical Notation
… unless the Systems Biology Graphical Notation is used.
SBGN webpage: http://sbgn.github.io/sbgn/
standardized glyphs and arcs and semantics specific to each.
sinksource
Process DescriptionBiochemistry
Activity FlowMolecular biology
Entity RelationshipPhysiology, Genetics
Three languages are defined.
SBGN webpage: http://sbgn.github.io/sbgn/
9
Objective: standardize diffs visualization
Highlighting of the differences between two model versions using SBGN.
RAF
ERK
MEK MEK
2P
ERK
MEK
P
ERK
P
RAF
P
ERK
2P
ADPATP
ATP ADP
ATP
ATP ADP
ADP
Version 1
MEK
2P
RAF
P
ERK
P
RAS
active
ERKERK
2P
ERK
RAS
MEK
P
RAF
ADP
ADP
ATP ATP
ATP
ADPATP
ADP
Version 2
10
Objective: standardize diffs visualization
Highlighting of the differences between two model versions using SBGN.
MEK
2P
RAS
ERK
RAF
P
RAF
RAS
active
ERK
2P
ERK
MEK
P
ERK
P
MEK
ATP
ATP
ATP ADP
ADP ADP
ADP
ATPADP
ATP
RAF
ERK
MEK MEK
2P
ERK
MEK
P
ERK
P
RAF
P
ERK
2P
ADPATP
ATP ADP
ATP
ATP ADP
ADP
Version 1
MEK
2P
RAF
P
ERK
P
RAS
active
ERKERK
2P
ERK
RAS
MEK
P
RAF
ADP
ADP
ATP ATP
ATP
ADPATP
ADP
Version 2Transition version
11
What are the steps?
Scharm et al. (2015) An algorithm to detect and communicate the differences in computational models describing biological systems - Bioinformatics.
1) Generate a tool to compare differences between two versions
The description of a model’s evolution is performed by the BiVeS library.
Biochemical Model Version Control System
http://bives.sems.uni-rostock.de
C
D
H E
com
mun
icatio
n pb
A
B
C D E
F
G
A
B
D H E
F
Gev
alua
tion p
b
A
B
C D E
F
G
A
B
D H E
F
Gprop
agat
ion p
b
A
B
C D E
F
G
A
B
D H E
F
Gid=“species1” id=“species1”
initi
alm
appi
ng p
b
A
B
C D E
F
G
A
B
D H E
F
G
model version 1model version 2
list of species list of reactions
C + D E D + H E
pre-
proc
essin
g pb
pre-
proc
essi
ngco
mm
un
ica
tion
eval
uatio
npr
opag
atio
nin
itial
map
ping
12
What are the steps?
Scharm et al. (2015) An algorithm to detect and communicate the differences in computational models describing biological systems - Bioinformatics.
1) Generate a tool to compare differences between two versions2) Generate a visual of the differences
Differences are colour coded:
http://budhat.sems.uni-rostock.de
U. Liebal - “Proteolysis of beta-galactosidase following SigmaB activation in Bacillus subtilis” (Mol. BioSyst)Model Versions: laczsynth-2012-11-10 and laczsynth-2012-11-11
13
What are the steps?
1) Generate a tool to compare differences between two versions2) Generate a visual of the differences3) Make the visual SBGN-compliant
Differences VisuaLization
https://github.com/Gebbi8/DiVil
U. Liebal - “Proteolysis of beta-galactosidase following SigmaB activation in Bacillus subtilis” (Mol. BioSyst)Model Versions: laczsynth-2012-11-10 and laczsynth-2012-11-11
14
Preliminary workflow
BiVeSBioModels
CellML
highlighted network
SBGN-MLPNG
JPEG
SVG
m odels v ersionsv ersion
transition
d ifferences
visualization
library
compute
d ifferences
s pecies B
s pecies A
1
2 3 4
5
89
10
11 12
highlighted
reaction
network
6
report of
changes
diff encoded in
XML
SBGN PD
compliant JSON
DOT language
Graph-ML
JSON
7
15
Preliminary workflow
BiVeSBioModels
CellML
highlighted network
SBGN-MLPNG
JPEG
SVG
m odels v ersionsv ersion
transition
d ifferences
visualization
library
compute
d ifferences
s pecies B
s pecies A
1
2 3 4
5
89
10
11 12
highlighted
reaction
network
6
report of
changes
diff encoded in
XML
SBGN PD
compliant JSON
DOT language
Graph-ML
JSON
7
16
Preliminary workflow
BiVeSBioModels
CellML
highlighted network
SBGN-MLPNG
JPEG
SVG
m odels v ersionsv ersion
transition
d ifferences
visualization
library
compute
d ifferences
s pecies B
s pecies A
1
2 3 4
5
89
10
11 12
highlighted
reaction
network
6
report of
changes
diff encoded in
XML
SBGN PD
compliant JSON
DOT language
Graph-ML
JSON
7
17
Preliminary workflow
BiVeSBioModels
CellML
highlighted network
SBGN-MLPNG
JPEG
SVG
m odels v ersionsv ersion
transition
d ifferences
visualization
library
compute
d ifferences
s pecies B
s pecies A
1
2 3 4
5
89
10
11 12
highlighted
reaction
network
6
report of
changes
diff encoded in
XML
SBGN PD
compliant JSON
DOT language
Graph-ML
JSON
7
18
Preliminary workflow
BiVeSBioModels
CellML
highlighted network
SBGN-MLPNG
JPEG
SVG
m odels v ersionsv ersion
transition
d ifferences
visualization
library
compute
d ifferences
s pecies B
s pecies A
1
2 3 4
5
89
10
11 12
highlighted
reaction
network
6
report of
changes
diff encoded in
XML
SBGN PD
compliant JSON
DOT language
Graph-ML
JSON
7
19
Preliminary workflow
BiVeSBioModels
CellML
highlighted network
SBGN-MLPNG
JPEG
SVG
m odels v ersionsv ersion
transition
d ifferences
visualization
library
compute
d ifferences
s pecies B
s pecies A
1
2 3 4
5
89
10
11 12
highlighted
reaction
network
6
report of
changes
diff encoded in
XML
SBGN PD
compliant JSON
DOT language
Graph-ML
JSON
7
20
Importance of SBO terms
without SBO terms with SBO terms
cell
u
z
cell
u
EmptySet
z
21
Importance of SBO terms
without SBO terms with SBO terms
cell
u
z
cell
u
EmptySet
z
Annotate your models !
22
Importance of SBO terms
without SBO terms with SBO terms
cell
u
z
cell
u
EmptySet
z
SBGN workshop: still time to join us!
Annotate your models !
23
Conclusion
Necessary improvements:● The SBGN output● Colour information● User interface
Opportunity for a student internship in Rostock (2-4 months project)More information: www.sems.uni-rostock.de
Prototype: SBGN map showing differences between model versions
● Extension of BiVeS library → information for generating SBGN map● Automatic force-based layout● Generation of an SBGN-ML file and graphical exports
24
Thank you for your attention!
Contact: [email protected]
Martin ScharmBiVeS and Budhat
Tom GebhardtDiViL
The SEMS team
Funded by