Tutorial session 3 Network analysis Exploring PPI networks using Cytoscape EMBO Practical Course Session 8 Nadezhda Doncheva and Piet Molenaar
Overview Focus: Network analysis Identify active subnetworks Analyze Gene Ontology enrichment Perform topological analysis Find network clusters Find network motifs Concepts Enrichment Clustering Guild by association Data Stored sessions; Drosophila and Neuroblastoma 12/6/20152
Identify active subnetworks jActiveModules plugin Active modules are sub-networks that show differential expression over user-specified conditions or time-points Microarray gene-expression attributes Mass-spectrometry protein abundance Input: interaction network and p-values for gene expression values over several conditions Output: significant sub-networks that show differential expression over one or several conditions 12/6/20153
jActiveModules (Demo) 12/6/20154
Use case; Assignment 3.1 Using neuroblastoma cell lines inhibitors to elucidate important pathways 2 neuroblastoma cell lines: SHEP21, SY5Y 7 inhibitors Profiled on Affymetrix array Other resource e.g. GEO 12/6/20155
Use case; Assignment 3.1 Systematic perturbations Different cell-lines Including controls: DMSO 97 arrays: data collected from R2: hugo-once etc PI3K signature RAS/ERK signature RAS/ERK-dependent Cell lines -SHEP2 -RD PI3K-dependent Cell lines -SY5Y -D425 Harvest: RNA Affy (97samples) protein WB PI3K AKT mTORC1 mTORC2 RAS MEK RAF ERK PI103 PP242 PIK90 Rapamycin U0126 AKTi 1/2 MK2206
Use case; Assignment Open the Neuroblastoma session and load the pvalues from this experiment 2. Run jActiveModules on the annotated network 1. What regions are important? 2. Can you imagine any caveats for this method? 12/6/20157
Assignment 3.1: results 1. Important regions 1. Several clusters; those with most mutations might deliver additional wet lab testable pathway players (drugtargets?) 2. Caveats: 1. Maintenance (housekeeping) processes 2. Known pathways only 12/6/20158
Gene Ontology Provides three structured controlled vocabularies (ontologies) of defined terms representing gene product properties: Biological Process (23074 terms): biological goal or objective Molecular Function (9392 terms): elemental activity/task Cellular Component (2994 terms): location or complex 12/6/20159
Analyze Gene Ontology enrichment BiNGO plugin: Calculates over-representation of a subset of genes with respect to a background set in a specific GO category Input: subnetwork or list, background set by user Output: tree with nodes color reflecting overrepresentation; also as lists Caveats: Gene identifiers must match; low GO term coverage, background determining 12/6/201510
BiNGO (Demo) 12/6/201511
Use case; Assignment Open the Neuroblastoma session and run BiNGO on the filtered network. 1. What categories are enriched? 2. Can you find these back in the article? 12/6/201512
Assignment 3.2: results 1. Quite some categories! 1. Filter out less informative top level categories: in several deeper categories neuron projection pops up 2. A clustering method can specify 3. Use subsets only 4. Worth mentioning: other tools eg. David 2. In second cluster neuron projection clearer; and large set of mutated genes 12/6/201513
Compute topological parameters NetworkAnalyzer plugin: inf.mpg.de/netanalyzer/ Computes a comprehensive set of simple and complex topological parameters Displays the results in charts, which can be saved as images or text files Can be combined with the ShortestPath plugin ndex.html 12/6/201514
NetworkAnalyzer (Demo) 12/6/201515
Identify hubs CytoHubba plugin: Computes several topological node parameters Identifies essential nodes based on their score and displays them in a ranked list Generates a subnetwork composed of the best-scored nodes 12/6/201516
CytoHubba (Demo) 12/6/201517
Use case; Assignment 3.3 Open the Drosophila network session 1. Check the network parameters 1. Is it scale free? 2. Can you find important players? 12/6/201518
Assignment 3.3: results 1. It is scalefree; the node degree distribution fits a power law 2. Depends on the type of player you want to find; between processes or master regulator over number of genes? 12/6/201519
Find network clusters Network clusters are highly interconnected sub-networks that may be also partly overlapping Clusters in a protein-protein interaction network have been shown to represent protein complexes and parts of biological pathways Clusters in a protein similarity network represent protein families Network clustering is available through the MCODE Cytoscape plugin: 12/6/201520
MCODE & ClusterMaker (Demo) 12/6/201521
Use case; Assignment 3.4 Open the Drosophila session 1. Run the MCODE algorithm 2. Run the MCL clustering algorithm 1. Compare the results 2. Can you corroborate some of the clusters found in the article? 3. Are there additional filtering options? 4. Play with the settings and observe their influence 12/6/201522
Assignment 3.4: results 1. MCODE gives fuzzier clusters 2. E.g. the syx-syb cluster 3. The cluster parameters are set as attributes; these can be used to filter 4. Less stringent settings will produce additional clusters, but also larger clusters 12/6/201523
Find network motifs NetMatch plugin: Network motif is a sub-network that occurs significantly more often than by chance alone Input: query and target networks, optional node/edge labels Output: topological query matches as subgraphs of target network Supports: subgraph matching, node/edge labels, label wildcards, approximate paths 12/6/201524
NetMatch (Demo) 12/6/201525
Use case; Assignment In the Drosophila session try to find a feedforward motif 2. Finally toy around with the settings of the Vizmapper to produce a nice paper-ready figure! 12/6/201526
Assignment 3.5: results 1. Simple feed forward gives lots of matches 1. Add attributes, or make more complex queries 2. Toying around can produce nice results! 12/6/201527
Other Useful Plugins PSICQUICUniversalClient AgilentLiteratureSearch GeneMANIA CyThesaurus structureViz ClusterMaker EnrichmentMap PiNGO ClueGO RandomNetworks 12/6/201528
Wrapping up… Biological questions I have a protein Function, characteristics from known interactions I have a list of proteins Shared features, connections I have data Derive causal networks Network Topology Hubs Clusters 12/6/ New hypotheses
End! And a final note….. 12/6/201530
Announcing Cytoscape 3.0 Beta Easier data import Improved user experience Graphical annotations One-click install from AppStore Improved API for app developers