The European Nutrigenomics Organisation Using pathway information to understand omics data Chris Evelo NuGO WP7 BiGCaT Bioinformatics Maastricht.

the European Nutrigenomics Organisation

NuNuGOGONuNuGOGO

Using pathway information to understand omics data

Chris EveloNuGO WP7

BiGCaT Bioinformatics Maastricht


NuNuGOGONuNuGOGOUn Oslo

Un Munich

Un Florence

Un Balearic Illes

Un Cork

Trinity

Un. Ulster

Rowett

Un Newcastle

Un Reading

IFR DiFE

Un Krakow

Inserm Marseille

TNO

Un Wageningen

Un Maastricht

EBI

NuNuGOGO

Un Lund

RikiltRivm


NuNuGOGONuNuGOGO Understanding Array data

Typical procedure

1. Annotate the reporters with something useful (UniProt!)

2. Sort based on fold change

3. Search for your favorite genes/proteins

4. Throw away 95% of the array


NuNuGOGONuNuGOGO



Typical procedure

1. Annotate the reporters with something useful (UniProt!)

2. Sort based on fold change

3. Search for your favorite genes/proteins

4. Throw away 95% of the array



“Advanced” procedures

o Gene clustering or principal component analysis

o Get groups of genes with parallel expression patterns

o Useful for diagnosis

o Not adding much to understanding (unless combined)


NuNuGOGONuNuGOGO Mapping

Annotation/coupling


NuNuGOGONuNuGOGOBest known: GenMAPP

Free, academic initiative with editable mapps,

collaborates with NuGO


NuNuGOGONuNuGOGO Best known: GenMAPP

• Full content of GO database

• Textbook like local mapps

• Geneboxes with active backpages, coupled to online databases

• Visualize anything numerical(fold changes on arrays, p-values, present calls, proteomics results)

• Update mapps yourself


NuNuGOGONuNuGOGO GenMAPP: Full GO content


NuNuGOGONuNuGOGO

GenMAPP:Textbook like maps

Extensive backpages

present with links to online

databases


NuNuGOGONuNuGOGO2D gels of 3T3-L1 (pre)-adipocytes

Enlarged sections

gels derived from:

A: 3T3-L1 pre-adipocytes,

B: 3T3-L1 adipocytes,

C: 3T3-L1 adipocytes with caloric restriction

D: 3T3-L1 adipocytes with caloric restriction and

TNF-a.


NuNuGOGONuNuGOGO

GenMAPP: visualize anything numerical

Example

Proteomics results (2D gels with GC-MS identification).

Fasting/feeding study shows regulation of glycolysis (data from Johan Renes, UM).

Other useful things:

- p-values, present calls- presence in clusters- presence in QTLs


NuNuGOGONuNuGOGO Update mapps yourself

You can do anything.E.g. add genes, annotation, backpage information, graphics

Next page shows a combination of metabolic mapps.

“The Nutrigenomics Masterpiece”

created by Milka Sokolović (AMC Amsterdam)


NuNuGOGONuNuGOGO MAPPfinder

• Ranks mapps where relatively many changes occur

• Useful to find unexpected pathways

• Statistics hardly developed


NuNuGOGONuNuGOGO MAPPfinder z-score

Number of genes/proteins changed on this

mapp

Expected number of changes

Standard deviation of

observed number

many dependencies to overcome


NuNuGOGONuNuGOGO MAPPfinder

• Next example from heart failure study(Schroen et al. Circ Res; 2004 95: 506-514)


NuNuGOGONuNuGOGO GenMAPP: Full GO content


NuNuGOGONuNuGOGO Scientist know GenMapp

Advantages: • Free,

• Runs on (high end) MS Windows,

• Relatively easy to use,

• Reasonable visualization,

• Some pathway statistics,

• Interesting content (Including GO, KEGG),

• Content editable,

• Adopting standards (e.g. BioPax),

• Soon to become open source.


NuNuGOGONuNuGOGO Scientist know GenMapp

Disadvantages: • Small academic initiative, uncertain lifespan

• No info on reactions, metabolites, location

• No change (e.g. time course) visualization

• Hard to cope with ambiguous reporters(we are working on that)

• Content could be better!


NuNuGOGONuNuGOGOMetacore example

GeneGo, IncGeneGo, Inc

• Systems ReconstructionTM Technology

www.genego.com


NuNuGOGONuNuGOGO

Agilent Affymetrix Proteomic SAGE

Concurrent visualization of different data types


NuNuGOGONuNuGOGO

GeneGo: primitive view of multiple conditions

Can you really see what happens?


NuNuGOGONuNuGOGOBuild new networkusing MetacoreTM from GeneGO

• Around p53 protein• Making us of biological DB• Filtered to reduce complexity:

– for ‘rat ortholog’– for ‘transcriptional regulation’– for ‘liver’


NuNuGOGONuNuGOGO


NuNuGOGONuNuGOGO

Filtering needed to reduce complexity


NuNuGOGONuNuGOGO


NuNuGOGONuNuGOGO Datasources 1

GenMAPP local MAPPs:

Largely created by a single postdoc (Dr.Kam Dahlquist).



KEGG:

Older pathway database (Kyoto Japan), on enzyme code (EC) level.

Example… The Homo Sapiens Urea cycle Mapp

A converted KEGG Mapp

Note that not all EC’s were converted and that they don’t have backpages.



KEGG Conversion:

= How would you convert EC codes to Swissprot codes?

1) Go to Swissprot, look for EC code

2) Add all proteins with that EC code to GenMapp backpage

Example: Superoxide dismutase function reaction would have:Cu/Zn-SOD, Mn-SOD and Ex-SOD in backpage… (and that is not what we usually want.

Note that many other tools use KEGG converted pathways (e.g. Spotfire Decissionsite, GeneGo, Ingenuity)



KEGG:

Another example: Apoptosis KEGG Mapp

A contributed Mapp

Somebody manually converted this Mapp!

Great work… But, there are only four of these



Gene Ontology Database:

Simple tree structure database with a of lot biological content (biologist know and like it).

Automatic annotation possible even for EST’s

See structure in MappFinder (1) (or use Go browser)



Alternative programs like GeneGo:

Based on expert knowledge (20 Russian biochemists).


NuNuGOGONuNuGOGO

NuGO data pathway data collection workflow

Combine and forwardexisting maps

to limited group of experts

Text miningfrom key genes/metabolites

Forward improved mapsto limited group of experts

Collect back page info

Forward new draft to alarger group of experts

within NuGO

Develop storage format plus tools

Think of best way to storepathway information

Develop/adapt entry toolsplus converters

Test resulting maps

Make maps available


NuNuGOGONuNuGOGO

BioPAX Plus/GMML 2

Working with Reactome, GenMapp and BioPax

Expert data

Reactome

BioPAX

GMML

Current GenMapp

GenMapp 2

NUGO/EBI

EBIMDP4/GenMapp

With Philippe Rocca and Imre

Vastrik (EBI/Reactome) we will define a way to

get Reactome views and export

them to GenMapp2

BiGCaT students created GenMapp 2 – GMML converters

with help from Lynn Ferrante (GenMapp.org)

Rachel van Haaften (BiGCaT/NuGO) and

Marjan van Erk (TNO/NuGO) visited EBI early 2005 to learn doing this

This step has not been taken care off as of yet…

Rachel van Haaften (BiGCaT/NuGO) and

Marjan van Erk (TNO/NuGO) will

test this and give user feedback

GMML (GenMapp Markup Language) is a superset of BioPAX

1. BioPAX could contain graphical views. (GMML 2 =

BioPAX2).

But, how doe we make that happen?


NuNuGOGONuNuGOGO Can it help you?

Seeing the errors

and getting useful information

A NuGO example

Red Wine Polyphenols (Dr Cristina Luceri)


NuNuGOGONuNuGOGOClusters in control grouprepresenting pathways

Hierarchical Clustering

7.12E3 0

1 1706 2 2

rat 5 rat 12 rat 14 rat 13 rat 4 rat 3 rat 2 rat 1 rat 15 rat 11

Scatter Plot

column for spotfire0 10 20 30 40 50 60 70 80

-80

-70

-60

-50

-40

-30

-20

-10

0

Caused by bad

technology and bad design


NuNuGOGONuNuGOGOAfter adapted normalization:


NuNuGOGONuNuGOGO The bioinformatics

BiGCaT Bioinformatics• Chris Evelo• Rachel van Haaften• Arie van Erk• Stan Gaj• Magali Jaillard• Kitty ter Stege• Thomasz Kelder• Gijs Huisman

TNO Zeist• Rob Stierum• Marjan van Erk

EBI Hinxton• Susanna Sansone• Philippe Rocca• Imre Vastrik

University Firenze• Duccio Cavalieri

GenMAPP.org• Bruce Conklin• Lynn Ferrante


NuNuGOGONuNuGOGO The Biology

Proteomics• Johan Renes (UM)

• Chris Evelo (BiGCaT)

The masterpiece• Milka Sokolović (AMC)

• Wout Lamers (AMC)

• Magali Jaillard (BiGCaT)

Heart Failure• Blanche Schroen (UM)

• Yigal Pinto (UM)

• Arie van Erk (BiGCaT)

Red Wine Polyphenols• Cristina Luceri (Firenze)

• At BiGCaT!

RhoA Stolen from• Rob Stierum (TNO)

Financial contributions: UM, TUe, Senter IOP, WCFS/ICN, Dutch Heart Foundation, NuGO

The European Nutrigenomics Organisation Using pathway information to understand omics data Chris Evelo NuGO WP7 BiGCaT Bioinformatics Maastricht.

Documents

array slide

nugo slide

content slide

developed slide

combined slide

nutrigenomics masterpiece

online databases slide

lund rikilt rivm slide