www.reactome.org hinxton 30 march 2011 1 reactome functional interaction (fi) network steve jupe

www.reactome.org Hinxton 30 March 20111

Reactome Functional Interaction (FI) Network

Steve Jupe

www.reactome.org Hinxton 30 March 20112

Motivation for Pathway and Network Analysis in Cancer

• “Exclusivity principle” in cancer.

• once a gene is mutated in a pathway, the selective growth advantage incurred by a second mutated gene in the same pathway is minimal.

• rare for a single tumor to have multiple genes altered in the same pathway.

• Analyzing mutated genes in a pathway context.

• reveals relationships among these genes.

• facilitates hypothesis generation on roles of these genes in disease phenotype.

• identifies genes on the affected pathways as therapeutic targets.

www.reactome.org Hinxton 30 March 20113

Analysis Workflow – Cancer Microarray

Experiment: Gene expression profiling of pancreatic tumour (vs normal tissue).

Output: list of genes that show significant changes in expression.

Downstream analysis: 1. Validation of the top genes sorted by fold-change (or p-value):

artificial cut-off, information loss;

2. Manual reviewing and gene classification based on literature, databases and own prior experience: time-consuming and bias;

3. GO annotation and pathway analysis (e.g. Reactome): what biological terms or functional categories are represented in the gene list more often than expected by chance;

4. Network-based analysis. - identify markers not as individual genes but as sub-networks extracted from protein interaction databases and the resulting sub-networks provide novel hypotheses for pathways involved in cancer.

www.reactome.org Hinxton 30 March 20114

What is the Functional Interaction Network?

• A reliable interaction network based on manually curated pathways extended with verified interactions

• Covers almost 50% of human Uniprot proteins

• A resource for construction of FI sub-networks based on gene sets

• Tools that: • Provide the underlying evidence for FIs • Build and analyze network modules of highly-interacting groups • Perform functional enrichment to annotate modules • Expand the network by finding genes related to the experimental data • display source pathway diagrams and overlay with a variety of information

sources such as cancer gene index annotations.

• Method and practical application: A human functional protein interaction network and its application to cancer data analysis, Wu et al. 2010 Genome Biology

www.reactome.org Hinxton 30 March 20115

Pairwise Problem

Input1-Input2, Input1-Catalyst,Input1-Activator, Input1-Inhibitor, Input2-Catalyst, Input2-Activator, Input2-Inhibtior,Catalyst-Activator, Catalyst-Inhibitor, Activator-Inhibitor

Reaction Functional Interaction

• Convert reactions in pathways into pair-wise relationships

• Functional Interaction: an interaction in which two proteins are involved in the same reaction as input, catalyst, activator and/or inhibitor, or as components in a complex

www.reactome.org Hinxton 30 March 20116

Reactome Functional Interaction (FI) Network [15%]

Wu et al. Genome Biology 2010

Construction of the Reactome FI Network

www.reactome.org Hinxton 30 March 20117

Coverage after Prediction

• 10,956 proteins (9,542 genes)

• 209,988 FIs

• 5% of network shown here

www.reactome.org Hinxton 30 March 20118

Reactome Pathway Diagram with FI Overlay

www.reactome.org Hinxton 30 March 20119

Limitations of web FI overlay

• Simple viewer – no extra tools

• Not easy to produce networks

• Not dynamic – you can’t move the pathway diagram objects

www.reactome.org Hinxton 30 March 201110

Projecting Experimental Data onto FI Network

www.reactome.org Hinxton 30 March 201111

A FI Sub-Network Generated from Significant Network Modules based on a TCGA OV Expression Data Set

www.reactome.org Hinxton 30 March 201112

Cytoscape

• Cytoscape - open source platform for visualizing complex-networks and integrating these with any type of attribute data.

• Java-based

• Plug-in development framework

• First public release: July 2002, v0.8

• Current release: 2.8.1

• Website: http://www.cytoscape.org

www.reactome.org Hinxton 30 March 201113

www.reactome.org Hinxton 30 March 201114

Reactome FI Cytoscape Plug-In

• Cytoscape v2.7.0 or later.

• Reactome Homepage (see URL below)

• Menu Bar -> Tools, FI Cytoscape Plugin

• Save jar file caBigR3.jar into Cytoscape plugins folder, restart.

• http://reactomedev.oicr.on.ca:8080/caBigR3WebApp/Cytoscape/plugins/caBigR3.jar

• Also available as Java Web Start: Cytoscape.jnlp

• http://reactomedev.oicr.on.ca:8080/caBigR3WebApp/Cytoscape/Cytoscape.jnlp

www.reactome.org Hinxton 30 March 201115

FI Network Analysis Pipeline

Select gene list (mutated, over-expressed, down-regulated, amplified or deleted genes)

Project genes of interest onto Reactome F.I. Network

Identify Disease/Cancer Subnetwork

Apply Clustering Algorithms

Apply Pathway/GO Annotation to each cluster

Generate Biological Hypothesis!Predict Disease Gene Function

Classify Patients & Samples

T2DOvCaGBM

BreastProstate

TCGAICGC

www.reactome.org Hinxton 30 March 201116

Using the Reactome FIs Plug-in

• Choose Plugins, Reactome FIs.

• FI plug-in supports three file formats:• Simple gene set: one line per gene.

• Gene/sample number pair. Contains two required columns, gene and number of samples having gene mutated, and an optional third column listing sample names (delimited by semi-colon ;).

• NCI MAF (mutation annotation file).

www.reactome.org Hinxton 30 March 201117

File Formats

MSI2PTPRTPELOSLC18A1TACC2FAM148BPRC1MSTNATP6V1G2APOEIMPA2AGERXPO5MESTRREB1BAT1WIPI1CATSPERBSSR1VEGFAEGFL8

Simple Gene ListGene/Sample Number PairsNCI MAF (mutation annotation file)

www.reactome.org Hinxton 30 March 201118

Loading a file

• Choose file

• Choose file format and parameters

• Click OK

www.reactome.org Hinxton 30 March 201119

Results Display

• Constructed network is displayed in the Network View panel using an FI specific visual style

www.reactome.org Hinxton 30 March 201120

Pop-up menu

• Right-click an empty space in the network view panel.

www.reactome.org Hinxton 30 March 201121

Fetch FI annotations

• Details information on selected FIs. Three edge attributes are created: FI Annotation, FI Direction, and FI Score. Edges display direction attribute values.

• --> for activating/catalyzing

• --| for inhibition

• solid line for complexes or inputs

• --- for predicted FIs.

• VizMapper tab, Edge Source Arrow Shape and Edge Target Arrow Shape values give details.

www.reactome.org Hinxton 30 March 201122

Analyze network functions

• Pathway or GO term enrichment analysis for the displayed network. • Use Data Panel to:

• Filter enrichment results by FDR value. • Check “Hide nodes in not selected rows” to display only nodes

in the selected row or rows • To analyze an unlinked set of genes choose the "Show genes

not linked to others" option when importing data

www.reactome.org Hinxton 30 March 201123

Cluster FI network

• Runs spectral partition based network clustering (Newman, 2006) on the displayed FI network.

• Nodes in different network modules will be shown in different colours (max 15 colours).

www.reactome.org Hinxton 30 March 201124

Analyze module functions

• Pathway or GO term enrichment analysis on individual network modules. • Use filter to remove small network modules

• Filter by FDR

• Button selects nodes in the network corresponding to selected row or rows in the Data Panel.

www.reactome.org Hinxton 30 March 201125

Other features - Query FI source

• Right-click edge, select "Reactome FI/Query FI Source".

• If the FI was extracted from a curated pathway the original data source(s) is displayed

• Double click a row in the table to show details

• If the FI is predicted the evidence is displayed.

www.reactome.org Hinxton 30 March 201126

Other features - Fetch FIs for node

• Select node in the network panel, right click for pop-up menu, select "Reactome FI, Fetch FIs"

• FI partners for the selected node are displayed in two sections: Partners that are displayed and partners that are not displayed.

• Select undisplayed partners and click Add

www.reactome.org Hinxton 30 March 201127

Other features - Show pathway diagram (KEGG)

• Select a pathway in "Pathways in Network/Modules" tabs, right click,

select "Show Pathway Diagram”

• KEGG pathways - diagram page is opened. Nodes listed in "Nodes"

column are highlighted in red.

www.reactome.org Hinxton 30 March 201128

Other features - Show pathway diagram (Reactome)• Reactome pathway diagrams open in a separate window.

• For human data the manually drawn pathway diagram is displayed.

• For other species an automated layout is generated.

• Nodes in the Cytoscape network are highlighted in blue on the diagram

• Detailed information is viewed using a pop-up menu item "View

Instance".

• Zoom slider bottom of the window.

• Overview window top-right corner

www.reactome.org Hinxton 30 March 201129

Cancer Gene Index annotations

• Cancer Gene Index annotations can be added in two ways:

• popup menu "Load Cancer Gene Index" when no object is selected (left)

• popup menu "Fetch Cancer Gene Index" for a selected node (right)

www.reactome.org Hinxton 30 March 201130

Load Cancer Gene Index

• This method loads the tree of NCI disease terms in the left panel.

• Select a disease term in the tree to select all diagram objects that have this annotation or one of its sub-terms.

www.reactome.org Hinxton 30 March 201131

Fetch Cancer Gene Index

• This method is for viewing detailed annotations for the selected gene or protein.

• Annotations are sortable by PubMed ID, Cancer type, status, and several other criteria.

www.reactome.org Hinxton 30 March 201132

Cancer Gene Survival Analysis

• Based on a server-side R script that runs

either coxph or Kaplan-Meyer survival

analysis.

• Requires appropriate data file (see online).

• Popup menu “Analyze Module Functions,

Survival Analysis...”

• Results are displayed with a tab "Survival

Analysis" (right).

• You can display multiple analyses.

• Only Kaplan-Meyer analysis will show a plot

file. You can click the file to view the actual

plot (far right).

www.reactome.org Hinxton 30 March 201133

Summary

• Reactome is a highly reliable, human-curated database of biological pathways

and reactions.

• Expanded the human curated Reactome pathways by importing pathways from

other databases and integrating predicted FIs: ~50% coverage.

• Functional interaction network provides a powerful way to visualize cancer and

other disease genes to reveal functional relationships among these genes.

www.reactome.org Hinxton 30 March 201134

Future

• Implement multiple network clustering algorithms• Currently spectral partition only

• Future: edge-betweenness, MCL, and hierarchical clustering

• Integrate with other data types: gene expressions, CNVs…

• Seamless integration with Reactome pathway diagrams• Expand known cancer pathways by using FIs

• Integrate miRNA/target interactions to study miRNA activities on cancers

www.reactome.org Hinxton 30 March 201135

Software Demo