Overview PPI Related Concepts PPI Prediction and Verification - - PowerPoint PPT Presentation

RobinViz: Reliability Oriented

Bioinformatic Networks Visualization

http://code.google.com/p/robinviz/

• A. E. Aladağ, C. Erten, M. Sözdinler

Overview

• PPI Related Concepts
• PPI Prediction and Verification
• RobinViz
• Visualization Model
• Layout Algorithms
• Extra Features
• Final Remarks

PPI Networks: Proteins

• Proteins:
• Amino acid polymers linked with peptide bonds
• Control, mediate most processes in the cell:
• Enzymatic
• Transport
• Structural roles etc.

P38 (MAPK1) from PDB

• Proteins function via interactions

P38 interactions with other kinase activity proteins from RobinViz

PPI Networks: Interactions

PPI Networks: Prediction

• How to Predict Interactions?
• Experimental:
• AP-MS
• SDS-PAGE
• Y2H
• ........

From Junker at al. '08

• Computational:
• Gene Order
• Coevolutionary Profiling
• Coexpression Analysis

PPI Networks: Verification

• Problems
• Many False Positives/Negatives
• 80K Yeast protein interactions in all types
• Only 3% agreement by more than one type
• How to Verify Interactions?
• Co-function
• Co-localization
• Co-expression
• Reliability Assignment
• Computational methods using
• Verification concepts
• Orthology

PPI Networks: Visualization

• Graph Visualization
• Issues:
• Very large graphs
• Thousands of nodes
• Tens of thousands of interactions
• Models for integration of
• Interaction verification
• Interaction reliabilities
• Previous Tools:
• Cytoscape [Shannon et al. 03']
• GenePro [Vlasblom et al. '06]
• SpringScape [Ebbels et al. '06]
• Lack verification integration and/or reliabilities
• No real-time data retrieval from major databases

RobinViz: Underlying PPI Network

• Edge-weighted, undirected graph
• Construct SQLite: Naming
• User-specified multiple PPI:
• Organism, Experiment type
• Retrieve from BioGrid
• Unify naming and merge
• Interaction Reliabilities:
• Specified Organisms
• Retrieve from HitPredict
• Unify naming and merge

RobinViz: Visualization Model

• Underlying PPI network, G = (V, E)
• Two-level visualization
• Central view graph, Gc = (Vc, Ec)
• Node/Edge weighted
• u in Vc subset of V
• (u,v) in Ec union of edges from E
• Peripheral view graphs
• Edge weighted
• Subgraph of G induced by u in Vc
• Determine
• VC and weights of nodes in Vc
• Mapping M: Vc →P(V), where P power set

RobinViz: Visualization Model

RobinViz: Central View

• Co-ontology
• Vc: (User) GO categories from AmiGO
• M: (User) Annotation repositories from GO
• Each repository:
• genei → catx, caty, catz, ...
• Convert:
• catx → genei, genej, genek
• Filter with Vc, unify names, merge sources
• Node Weights:
• PPI hit ratio

RobinViz: Central View

• Co-expression
• Vc:
• (User) Expression matrix from GEO
• (User) Biclustering algorithm:
• CC, BIMAX, REAL
• Set of genes in each resulting bicluster
• M: Same as co-ontology
• Node Weights:
• PPI hit ratio
• H-value
• Functional enrichment

RobinViz: Bidirectional Verification

• From trustworthy central clusters to mistrusted PPI

False Positives False Negatives

RobinViz: Bidirectional Verification

• From trustworthy PPI to mistrusted central clusters

False Negatives False Positives

RobinViz: Detailed Analysis

• Co-ontology
• Detailed online information via AmiGO
• GO Table within the system

RobinViz: Detailed Analysis

• Co-expression
• GO Table, Functional Enrichment Table
• Heatmap and Parallel Plots for biclustering results

RobinViz: Layout Algorithms

• Spring Embedder Layout (Weighted)

RobinViz: Layout Algorithms

• Spring Embedder Layout
• Energy based: Nodes connected with springs
• Heuristic for minimizing energy of the system
• Edges short, non-edges long, symmetric layout
• Random initial positions
• For k iterations
• For each edge (u,v)
• Displace u, v proportional to Fattr between u, v
• For each node u
• For each node v
• Displace u proportional to Frep between u,v
• Straight-line edges for each edge (u,v)
• Running time Θ(|V|2k). In practice Θ(|V|3)
• Reliabilities: Weighted modification in force formula

RobinViz: Layout Algorithms

• Sugiyama-style Layout (Weighted)

RobinViz: Layout Algorithms

• Sugiyama-style Layout (Weighted)
• Layer assignment:
• Modify Coffman-Graham '72
• Longest path with promotion heuristic
• Minimizing weighted edge lengths, drawing area
• Order within layers
• Layer-by-layer sweep
• Weighted crossing minimization
• Coordinate assignment
• x-coord: 2-bends with Brandes et al. '02.
• y-coord: Proportional to density of weighted edges

RobinViz: Other Features

• Node Coloring
• 1st level in: Process, Compartment, Function, All
• Peripheral (Func) Central (Func) Central (All)
• Detailed 1-hop/2-hop neighborhoods
• Zooming/Animation/Selection Focus
• Search Panel
• Save/Load Session, Automatic Database Update

RobinViz: Libraries, Databases etc.

• Code and Libraries
• Graph Layout: C++ LEDA
• GUI and Data Processing: Python and PyQt4
• Settings Files: YAML
• Website Parsing: BeautifulSoup
• ~30.000 lines of code
• Windows and Linux versions
• Databases
• PPI Networks: BioGrid
• Interaction Reliabilities: HitPredict
• GO Tree: GeneOntology.org TermDB
• GO Associations: GeneOntology.org Association
• Gene Expression: GEO (NCBI)
• Web Services
• BioGrid.org and AmiGO

Future Work

• RobinViz Extensions:
• Hypothetical network creation: Nomenclature
• Embedded Network Analysis
• Hierarchical Clustering
• Integration of other popular databases
• Graph unions, new definitions
• PPI network prediction from multiple species