Improving the management of computational models

SYSTEMS BIOLOGY

BIOINFORMATICS

ROSTOCKS E Ssimulation experiment management system

Improving the Management ofComputational Modelsstorage retrieval & ranking version control

MARTIN SCHARM1,2 , RON HENKEL1 , DAGMAR WALTEMATH1 ,OLAF WOLKENHAUER1,31Department of Systems Biology & Bioinformatics, University of Rostock2Department of Computer Science, University of Oxford3Stellenbosch Institute for Advanced Study, South Africa

http://sems.uni-rostock.de

European Bioinformatics InstituteUK 2013

Dec 4, 2013 SEMS | Martin Scharm, Ron Henkel, Dagmar Waltemath, Olaf Wolkenhauer 1

http://www.sbi.uni-rostock.de/team/single/martin-scharm/http://www.sbi.uni-rostock.de/team/single/ron-henkel/http://www.sbi.uni-rostock.de/team/single/dagmar-waltemath/http://www.sbi.uni-rostock.de/team/single/prof-olaf-wolkenhauer/http://sems.uni-rostock.dehttp://www.sbi.uni-rostock.de/team/single/martin-scharm/http://www.sbi.uni-rostock.de/team/single/ron-henkel/http://www.sbi.uni-rostock.de/team/single/dagmar-waltemath/http://www.sbi.uni-rostock.de/team/single/prof-olaf-wolkenhauer/

Introduction

time

#mod

els

2013-06-18

462

+50

1+x1

03

216

+19

6

2009-06-1631

2005-04-11

Number of models is steadily increasing Models tend to get more complex Continuous development


http://www.sbi.uni-rostock.de/team/single/martin-scharm/http://www.sbi.uni-rostock.de/team/single/ron-henkel/http://www.sbi.uni-rostock.de/team/single/dagmar-waltemath/http://www.sbi.uni-rostock.de/team/single/prof-olaf-wolkenhauer/

Introduction

reac

tions

mod

el

~27.

06

curated

~288

.91

noncurated

2013 Number of models is steadily increasing Models tend to get more complex Continuous development



Introduction

time

#Ver

sions

Number of models is steadily increasing Models tend to get more complex Continuous development



IntroductionTasks

Storage & Availability

Search & Retrieval

Version Control



IntroductionTasks


Search & Retrieval

Version Control?



IntroductionTasks


Search & Retrieval

Version Controlversion x-1 version x version x+1



IntroductionSolutions

Several projects developmethods to support thecommunity

ModelRepository



SYSTEMS BIOLOGY

BIOINFORMATICS


track development

store retrieve

rank

Retrieval

Ranking

Version 1

Version 2

latest

Version Control

DocumentTyson1991 Cell

Cycle 6

varC2 pM Cell

Reaction3

CP

Uniprot:P04551

Uniprot:P04551

GO:0005623

Interpro

: IPR006670

isVe

rsio

nOf

isVe

rsio

n

hasP

art

is

Pubmed:1831270

Kegg Pathway

sce04111

isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

SEDML

Modelreference

Output

Datagenerator

Simulation Task

Variable

Variable

Document

Tyson_1991

C2 CP

time

environmen

t

isDescribedByPubmed:

1831270

time timeCPC2 CP C2


Cycle 6

varC2 pM Cell

Reaction3

CP

Uniprot:P04551

Uniprot:P04551

GO:0005623

Interpro

: IPR006670

isVe

rsio

nOf

isVe

rsio

n

hasP

art

is

Pubmed:1831270

Kegg Pathway

sce04111

isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf


Cycle 6

varC2 pM Cell

Reaction3

CP

Uniprot:P04551

Uniprot:P04551

GO:0005623

Interpro

: IPR006670

isVe

rsio

nOf

isVe

rsio

n

hasP

art

is

Pubmed:1831270

Kegg Pathway

sce04111

isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf


Cycle 6

varC2 pM Cell

Reaction3

CP

Uniprot:P04551

Uniprot:P04551

GO:0005623

Interpro

: IPR006670

isVe

rsio

nOf

isVe

rsio

n

hasP

art

is

Pubmed:1831270

Kegg Pathway

sce04111

isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Tyson_1991

C2 CP

time

environmen

t


1831270

time timeCPC2 CP C2


Cycle 6

varC2 pM Cell

Reaction3

CP

Uniprot:P04551

Uniprot:P04551

GO:0005623

Interpro

: IPR006670

isVe

rsio

nOf

isVe

rsio

n

hasP

art

is

Pubmed:1831270

Kegg Pathway

sce04111

isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf


Cycle 6

varC2 pM Cell

Reaction3

CP

Uniprot:P04551

Uniprot:P04551

GO:0005623

Interpro

: IPR006670

isVe

rsio

nOf

isVe

rsio

n

hasP

art

is

Pubmed:1831270

Kegg Pathway

sce04111

isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Tyson_1991

C2 CP

time

environmen

t


1831270

time timeCPC2 CP C2

Graph Database

Document

SEDML

Modelreference

Output

Simulation

Task

Document

SEDML

Modelreference

Output

Datagenerator

Simulation Task

Variable

Variable

Document

SEDML

Modelreference

Output

Datagenerator

Simulation Task

http://sems.uni-rostock.de/


http://sems.uni-rostock.de/http://www.sbi.uni-rostock.de/team/single/martin-scharm/http://www.sbi.uni-rostock.de/team/single/ron-henkel/http://www.sbi.uni-rostock.de/team/single/dagmar-waltemath/http://www.sbi.uni-rostock.de/team/single/prof-olaf-wolkenhauer/

StorageGraphDB

Taken from: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.174.8830&rep=rep1&type=pdf



StorageAvailable Data

Model & Simulation Annotation & Ontologies A models network

Constituent names XML encoded

representation

Biochemicalbackground

Synonyms

Model structure Aggregation values Model related data



StorageGraphDB

Document

Tyson1991 Cell Cycle 6

var

C2 pM CellReaction3 CP

Uniprot:P04551 Uniprot:P04551 GO:0005623Interpro: IPR006670

isVe

rsio

nOf

isVe

rsio

n

hasP

art

is

asProductasReactant isContainedIn

Pubmed:1831270

Kegg Pathwaysce04111

isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Tyson 1991 - SBML

Model

Entities

Annotations

fig. courtesy R. HenkelDec 4, 2013 SEMS | Martin Scharm, Ron Henkel, Dagmar Waltemath, Olaf Wolkenhauer 8


StorageGraphDB

Document

Model

sodium channel

sodium channel m

gate

time

envmt

has_annotation Pubmed:12991237

time timevm v m

is_connected is_connected

is_mapped_to

Tyson 1991 - CellML

Model

Component

Variable



StorageGraphDB

Document

SEDML

Modelreference Output

DatageneratorSimulation Task

Variable

Variable

Tyson 1991 - SED-ML

Simulation

Entity

Variable



StorageGraphDB

SBO:Ontology

SBO:0000

SBO:544 SBO:236SBO:231

isA

SBO:064 SBO:545SBO:004 SBO:003

Systems Biology Ontology

Root

1. Level

All Ontologies available in OWL can be added easily

fig. courtesy R. Henkel



StorageGraphDB

Models Model related data



StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf



StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Model

sodium channel

sodium channel m

gate

time

envmt


time timevm v m


is_mapped_to



StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Model

sodium channel

sodium channel m

gate

time

envmt


time timevm v m


is_mapped_to

Document

SEDML



Variable

Variable



StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Model

sodium channel

sodium channel m

gate

time

envmt


time timevm v m


is_mapped_to

Document

SEDML



Variable

Variable

SBO:Ontology

SBO:0000


isA




StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Model

sodium channel

sodium channel m

gate

time

envmt


time timevm v m


is_mapped_to

Document

SEDML



Variable

Variable

SBO:Ontology

SBO:0000


isA


Henkel et al. 2013, Graph-based storage of computational biology models and associated simulations, in preparation



StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Model

sodium channel

sodium channel m

gate

time

envmt


time timevm v m


is_mapped_to

Document

SEDML



Variable

Variable

SBO:Ontology

SBO:0000


isA


Index

Model

IdNameComponentVariableSpeciesReactionCompartment



StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Model

sodium channel

sodium channel m

gate

time

envmt


time timevm v m


is_mapped_to

Document

SEDML



Variable

Variable

SBO:Ontology

SBO:0000


isA


Index

Model

Publication

TitleJournalAbstractAuthors



StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Model

sodium channel

sodium channel m

gate

time

envmt


time timevm v m


is_mapped_to

Document

SEDML



Variable

Variable

SBO:Ontology

SBO:0000


isA


Index

Model

Publication

Person

First nameLast nameOrganizationEmail



StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Model

sodium channel

sodium channel m

gate

time

envmt


time timevm v m


is_mapped_to

Document

SEDML



Variable

Variable

SBO:Ontology

SBO:0000


isA


Index

Model

Publication

Person

Annotation

URIDescription



StorageGraphDB


Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Document

Model

sodium channel

sodium channel m

gate

time

envmt


time timevm v m


is_mapped_to

Document

SEDML



Variable

Variable

SBO:Ontology

SBO:0000


isA


Index

Model

Publication

Person

Annotation

Simulation

IdName



retrieval & rankingExample

Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Show me models byTyson describing the cellcycle and have cdc2!

Person

Annotation



retrieval & rankingExample

Document


var



isVe

rsio

nOf

isVe

rsio

n

hasP

art

is


Pubmed:1831270


isDescribedBy

is

EC-Code: 3.1.3.16

isVe

rsio

nOf

Thank you!

Person

Annotation

1. (0.859) Tyson1991 - Cell Cycle 6 var2. (0.854) Tyson2001_Cell_Cycle_Regulation3. (0.477) Chen2004 - Cell Cycle Regulation

Henkel et al. 2010: Ranked retrieval of Computational Biology modelsDec 4, 2013 SEMS | Martin Scharm, Ron Henkel, Dagmar Waltemath, Olaf Wolkenhauer 13


Version Control

version x-1 version x version x+1

what happened ?

track versions



Version Control

version x-1 version x version x+1

track versions

what happened ?



Model EvolutionExample: Cell Cycle

CyclinCdc2 P

CyclinCdc2 P

Modeling the cell division...

John J Tyson, 1991

muell




Cyclin

Cdc2 P

Cyclin

Cdc2 P

Cdc25Cdc25 Wee1 Wee1

Numerical analysis of a comprehensive model of M-phase control in Xenopus oocyte

Bela Novak and John J Tyson, 1993

Cyclin

Cdc2 P

Cyclin

Cdc2 P


John J Tyson, 1991

muell




Cyclin

Cdc2 P

Cyclin

Cdc2 P




Cyclin

Cdc2 P

Cyclin

Cdc2 P

Cdc25Cdc25

Mik1 Mik1

Wee1 Wee1

Quantitative analysis of a molecular model of mitotic control in Fission yeast


Cyclin

Cdc2 P

Cyclin

Cdc2 P


John J Tyson, 1991

muell




Cyclin

Cdc2 P

Cyclin

Cdc2 P




Cyclin

Cdc2 P

Cyclin

Cdc2 P

Cdc25Cdc25

Mik1 Mik1

Wee1 Wee1

Quantitative analysis of a molecular model of mitotic control in Fission yeast


Cyclin

Cdc2 P

Cyclin

Cdc2 P


John J Tyson, 1991

Cyclin

Cdc2 P

Cyclin

Cdc2 P

Cdc25Cdc25

Mik1 Mik1

Wee1 Wee1

Cyclin

Cdc2 P

Rum1

Modeling the control of DNA replication in fission yeast


muell



Version Controlgood news

A r C

B

D

cycE/cdk2

RB/E2F

RB-Hypo

free E2F

A r

B

C

D

E s

RB/E2F

RB-Hypo

free E2F

cycE/cdk2

RB-Phos

new insights

Waltemath et al.: Improving the reuse of computational models through versioncontrol. Bioinformatics (2013) 29(6): 742-728;


http://bioinformatics.oxfordjournals.org/content/29/6/742.longhttp://bioinformatics.oxfordjournals.org/content/29/6/742.longhttp://www.sbi.uni-rostock.de/team/single/martin-scharm/http://www.sbi.uni-rostock.de/team/single/ron-henkel/http://www.sbi.uni-rostock.de/team/single/dagmar-waltemath/http://www.sbi.uni-rostock.de/team/single/prof-olaf-wolkenhauer/

BiVeSDifference Detection

A r C

B

D

cycE/cdk2

RB/E2F

RB-Hypo

free E2F

A r

B

C

D

E s

RB/E2F

RB-Hypo

free E2F

cycE/cdk2

RB-Phos

A

r

BC

D

A

r

B

C

D

E

s

Biochemical Model Version Control System

compares models encoded in standadizedformats (currently: and )

maps hierarchically structured contentmapping



BiVeSDifference Detection

A r C

B

D

cycE/cdk2

RB/E2F

RB-Hypo

free E2F

A r

B

C

D

E s

RB/E2F

RB-Hypo

free E2F

cycE/cdk2

RB-Phos

A

r

BC

D

A

r

B

C

D

E

s

Biochemical Model Version Control System

compares models encoded in standadizedformats (currently: and )

maps hierarchically structured content constructs a diff (in XML format) is able to interprete this diff

Diff

movesproduct of r: C

deletesproduct of r: B

insertsspecies: Eproduct of r: Ereaction s

mapping

diff construction



BudHatDiff Visualization

A r C

B

D

cycE/cdk2

RB/E2F

RB-Hypo

free E2F

A r

B

C

D

E s

RB/E2F

RB-Hypo

free E2F

cycE/cdk2

RB-Phos

A

r

BC

D

A

r

B

C

D

E

s

Diff

movesproduct of r: C

deletesproduct of r: B

insertsspecies: Eproduct of r: Ereaction s

calls BiVeS to construct the diff displays the result in various formats

the XML diff a reaction network highlighting the

changes using a human readable report

A r B

C

D

E s



ApplicationsCommunicating Changes

Novak & Tyson 93: Numerical analysis of a comprehensive model of M-phase control in Xenopusoocyte extracts and intact embryos

Biomodels Database BIOMD0000000107:


http://www.ebi.ac.uk/biomodels-main/BIOMD0000000107http://www.sbi.uni-rostock.de/team/single/martin-scharm/http://www.sbi.uni-rostock.de/team/single/ron-henkel/http://www.sbi.uni-rostock.de/team/single/dagmar-waltemath/http://www.sbi.uni-rostock.de/team/single/prof-olaf-wolkenhauer/


Fixed the model How to communicate the changes?

Unix diff?

1c1< --->

287 lines...





BiVeS Diff?

much smarter, but still 24 lines..Dec 4, 2013 SEMS | Martin Scharm, Ron Henkel, Dagmar Waltemath, Olaf Wolkenhauer 21




BiVeS Diff: Highlighted chemical reaction network





BiVeS Diff: Report



ApplicationsBiomodelsDatabase / PMR2

Biomodels Database CellML Model Repository

963 # Models 158910193 # Versions 128789148 # prod. Diffs 12936

BiVeS Unix Diff BiVeS Unix Diff2233 1890 # empty Diffs 10113 10043

434.825 1977.210 #ops 349.5282 480.249

6915 non-empty diffs following 2823 non-empty diffs



ApplicationsBiomodelsDB / PMR2

Biomodels Database

0 20000 40000 60000 80000

050

000

1000

0015

0000

2000

0025

0000

3000

00

#operations needed by BiVeS

#ope

ratio

ns n

eede

d by

Uni

x' d

iff

#operations Unix' diff == #operations BiVeS

CellML Model Repository

0 5000 10000 15000 20000 25000

050

0010

000

1500

020

000

2500

0

#operations needed by BiVeS

#ope

ratio

ns n

eede

d by

Uni

x' d

iff

#operations Unix' diff == #operations BiVeS




upda

tes

inse

rts

1 5 10 50 100 500 1000

upda

tes

inse

rts

1 5 10 50 100 500 1000

Biomodels Database


move

insert

delete

update

move

insert

delete

update




attr

node

text

1 5 10 50 100 500 1000

attr

node

text

1 5 10 50 100 500 1000

Biomodels Database




Summary

SEMS develops methods for improved model management Storage

Retrieval & Ranking

Version Control

... and implements these methods in tools that are modularly designed

open source

easy to integrate



Future

Support for further standards PharmML

CombineArchive

SBGN

respect further ontologies in storage/search

Improved version control mapping w/ respect to biological meaning

scoring schemes for differences

filter for changes

Improved availability and analyzability of data



SYSTEMS BIOLOGY

BIOINFORMATICS


Thank you for your attention!

SEMS group

Dagmar WaltemathRon HenkelMartin PetersMarkus WolfienRebekka AlmOlaf Wolkenhauer

@SemsProject

http://sems.uni-rostock.de


https://twitter.com/semsprojecthttp://sems.uni-rostock.dehttp://www.sbi.uni-rostock.de/team/single/martin-scharm/http://www.sbi.uni-rostock.de/team/single/ron-henkel/http://www.sbi.uni-rostock.de/team/single/dagmar-waltemath/http://www.sbi.uni-rostock.de/team/single/prof-olaf-wolkenhauer/

BiVeSIntegration

jvm network cmd

import de.unirostock.sems.bives.api.SBMLDiff;

[...]

SBMLDiff differ = new SBMLDiff (sbmlFileA, sbmlFileB);differ.mapTrees ();String graph = differ.getCRNGraphML ();

[...]



BiVeSIntegration

jvm network cmd

curl -d {"get":[

"documentType","xmlDiff"

],"files":{

"versionA":"http://your.db/path/to/versionA.sbml","versionB":"http://your.db/path/to/versionA.sbml"

}} http://bives.server.tld



BiVeSIntegration

jvm network cmd

java -jar BiVeS.jar path/to/versionA path/to/versionB

git diff 88fea1cddf b64477d742 model.file



BiVeS -vs- Unix DiffArtifacts

delete a single line in an XML:

for Unix diff it is 1 operation. Bives detects 5 operations (delete node variableand 4 attributes).

Worse if multiple nodes on the same line:

Caffeine - Caffeine-induced Ca2+ transients from 18 individual myocytes

And worst if whole document only 1 line max. 2 operations for the Unix diff tool.



CombineArchive

Single file containing models and model related data:

simulation experiment descriptions in SED-ML graphical representations in SBGN-ML

necessary for the description of a model and all associated data and procedures simplifies exchange

Proposal by Nicolas Le Novre, Frank Bergmann, and Richard Adamshttp://co.mbine.org/documents/archive

We developed a JAVA library to create/read/manipulate CombineArchiveshttp://sems.uni-rostock.de/projects/combinearchive/

database import/export, difference detection between whole archives


http://co.mbine.org/documents/archivehttp://sems.uni-rostock.de/projects/combinearchive/http://www.sbi.uni-rostock.de/team/single/martin-scharm/http://www.sbi.uni-rostock.de/team/single/ron-henkel/http://www.sbi.uni-rostock.de/team/single/dagmar-waltemath/http://www.sbi.uni-rostock.de/team/single/prof-olaf-wolkenhauer/

BiVeSMapping

input: 2 XML documents map unambiguos

nodes/subtrees propagate mapping up/down exploit further knowledge of

biological systems

same ID

samesubtree

hash

...

... ...... ...

...

... ...... ...



BiVeSMapping



biological systems

eq. netw....

... ...... ...

...

... ...... ...



BiVeSMapping



biological systems

A B

C

D E

A B

C

D E

A B

C

D E



ApplicationsInvestigating BiomodelsDB / PMR2

bive

sun

ix

0 500 1000 1500

bive

sun

ix

0 500 1000 1500

Biomodels Database




ApplicationsInvestigating BiomodelsDB / PMR2

bive

sun

ix

0 500 1000 1500 2000 2500 3000 3500

bive

sun

ix

0 500 1000 1500 2000 2500 3000 3500

Biomodels Database




IntroductionSEMSGraphDBretrieval & rankingVersion ControlSummary & OutlookDoneAppendix