Docstoc

Protein Interaction Networks

Document Sample
Protein Interaction Networks Powered By Docstoc
					                               KOCSEA Technical Symposium 2010




Systematic Analysis of Interactome:
  A New Trend in Bioinformatics




           Young-Rae Cho, Ph.D.
              Assistant Professor
        Department of Computer Science
              Baylor University
History of Bioinformatics




     Stage 1.
 Sequence Analysis




 •   Gene sequencing
 •   Sequence alignment
 •   Homolog search
 •   Motif finding
History of Bioinformatics


              Computational Biology


     Stage 1.                   Stage 2.
 Sequence Analysis         Structure Analysis




 •   Gene sequencing       •   Protein folding
 •   Sequence alignment    •   Homolog search
 •   Homolog search        •   Binding site prediction
 •   Motif finding         •   Function prediction
History of Bioinformatics


              Computational Biology                      Functional Genomics


     Stage 1.                   Stage 2.                      Stage 3.
 Sequence Analysis         Structure Analysis            Expression Analysis




 •   Gene sequencing       •   Protein folding            •   Function prediction
 •   Sequence alignment    •   Homolog search             •   Gene clustering
 •   Homolog search        •   Binding site prediction    •   Sample classification
 •   Motif finding         •   Function prediction
History of Bioinformatics


              Computational Biology                      Functional Genomics              Systems Biology


     Stage 1.                   Stage 2.                      Stage 3.                        Stage 4.
 Sequence Analysis         Structure Analysis            Expression Analysis              Network Analysis




 •   Gene sequencing       •   Protein folding            •   Function prediction     •    Network modeling
 •   Sequence alignment    •   Homolog search             •   Gene clustering         •    Interaction prediction
 •   Homolog search        •   Binding site prediction    •   Sample classification   •    Function prediction
 •   Motif finding         •   Function prediction                                    •    Pathway identification
                                                                                      •    Module detection
Biological Networks

  Definition

      Maps of biochemical reactions, interactions, regulations between genes or proteins



  Importance

      Provide insights into the mechanisms of molecular function within a cell

      Significant resource for functional characterization of genes or proteins

      Require computational and systematic approaches



  Examples

      Metabolic networks

      Protein-protein interaction networks

      Genetic interaction networks

      Gene regulatory networks (Signal transduction networks)
Protein Interaction Networks

  Determination

      Experimental methods: Y2H, MS, Protein Microarray

      Computational methods: Homolog search, Gene fusion analysis, Phylogenetic profiles

      Genome-scale protein-protein interactions  Interactome



  Representation

      Un-weighted, undirected graph



  Challenges

      Unreliability

      Large scale

      Complex connectivity
Network Re-structuring

  Strategy                                                      unweighted network

      To resolve complex connectivity
                                                     edge weighting
      Converts the complex graph to

       a hierarchical tree structure                               weighted network

      Uses the concepts of path strength,    functional linkage measurement
       functional linkage, and centrality
                                                                   score matrix

  Process                                        network restructuring
      Input: a protein interaction network
                                                                  structured network
      Output: a list of functional modules
                                               hub confidence measurement

                                                                       hubs


                                                    network clustering

                                                                      clusters
Path Strength

  Path Strength Model

      Assumption: each node in a path chooses a succeeding edge based on the weighted

                      probability


     




  Path Strength Factors

      Edge weights

      Path length

      Node weighted degree
Functional Linkage

  Measurements

     Path strength of the strongest path between two nodes

     Computational problem

     Needs a heuristic approach

     Uses a user-specified threshold of the max path length




  Formula

     k-length path strength:


     Functional linkage:



                                             shortest path length   threshold
Network Restructuring

  Centrality

      Weighted closeness:



  Algorithm

      Computes centrality for each node a

      Selects a set of ancestor nodes, T(a), of a by

      Selects a parent node, p(a), of a by




  Example
Hub Confidence

  Measurement

     Selects a set of child nodes, D(a), of a by


     Selects a set of descendent nodes, La, of a by


     Computes the hub confidence, H(a), of a by




  Example
Clustering

  Algorithm

      Iteratively select a hub a with the highest hub confidence

      Output the sub-tree La including a as a cluster (functional module)



  Cluster Depth

      The max path length from the root of the sub-tree to a leaf




  Example
Topological Assessment of Hubs

  Network Vulnerability

      Random attack: repeatedly disrupt a randomly selected node

      Degree-based hub attack: repeatedly disrupt the highest degree node

      Structural hub attack: repeatedly disrupt the node with the highest hub confidence

      For each iteration, observes the largest component



  Results                                     1.00

                                               0.95
               fraction of largest component




                                               0.90

                                               0.85

                                               0.80

                                               0.75

                                               0.70            random attack
                                                               degree-based hub attack
                                               0.65
                                                               structural hub attack
                                               0.60
                                                      0   20      40    60    80    100   120   140   160

                                                                        number of nodes
Biological Assessment of Hubs

  Protein Lethality

      Determines lethal / viable proteins by knock-out experiment

      Lethality represents functional essentiality

      Orders proteins by degree and hub confidence

      Observes the cumulative proportion of lethal proteins for every 10 proteins



  Results                          1.0
                                                                structural hubs
                                    0.9                         degree-based hubs
                average lethality




                                    0.8


                                    0.7


                                    0.6


                                    0.5


                                    0.4
                                          0   20   40      60   80       100   120   140

                                                        number of hubs
Topological Assessment of Clusters

  Modularity

      A combined measure of density within each cluster and separability among clusters

      Estimated by the ratio of the number of edges within a cluster (sub-graph)

       to the number of all edges starting from the nodes in the cluster (sub-graph)

      Observes the average modularity of clusters with respect to the cluster depth



  Results

      More specific function module has                                  180
                                                                          160

       higher modularity




                                                     average modularity
                                                                          140

                                                                          120
      Justify the general-to-specific concepts                           100

                                                                          80
       of hierarchical functional modules                                 60

                                                                          40

                                                                          20
                                                                           0
                                                                                1   2   3   4    5   6   7   8   9   10 11 12

                                                                                                cluster depth
Biological Assessment of Clusters

  f-Measure

      Compares each output cluster X with the real functional annotation Y (from MIPS)

      Recall = (# of common proteins of X and Y) / (# of proteins in Y)

      Precision = (# of common proteins of X and Y) / (# of proteins in X)

      f-measure = 2 × Recall × Precision / (Recall + Precision)



  Results

      Compared with the results from previous hierarchical clustering methods, e.g.,

       edge-betweenness (top-down approach) and ProDistIn (bottom-up-approach)
Conclusion

  Motivation

      Significant functional knowledge in protein interaction networks (interactome)

      Complex connectivity



  Contributions

      Convert an unstructured network to a structured network

      Conserve functional information through pathways

      High network vulnerability, low functional lethality at hubs as a drug target

      Applicable to various fields, e.g., social networks, WWW

      Foundation of structural dynamics during network evolution
Questions ?

  Reference

     Y.-R. Cho and A. Zhang, “Identification of functional modules by converting

       interactome networks into hierarchical ordering of proteins”. BMC Bioinformatics,

       11(Suppl 3):S3, 2010

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:4
posted:3/13/2012
language:English
pages:19