html5-img
1 / 22

Gökay Burak AKKUŞ 2003700717

Assessing Experimentally Derived Interactions in a Small World Debra S. Goldberg, Frederick P. Roth Harvard Medical School. Gökay Burak AKKUŞ 2003700717. Agenda. Experimentally determined networks Small World networks Watts & Strogatz model Mutual Clustering Coefficients

Download Presentation

Gökay Burak AKKUŞ 2003700717

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Assessing Experimentally Derived Interactions in a Small WorldDebra S. Goldberg, Frederick P. RothHarvard Medical School Gökay Burak AKKUŞ 2003700717

  2. Agenda • Experimentally determined networks • Small World networks • Watts & Strogatz model • Mutual Clustering Coefficients • Protein-protein interaction • Predictions without direct experimental evidence • Conclusions

  3. Experimentally determined networks • “Experimentally derived networks are susceptible to errors” • True edges • False edges • From random graph • To regular lattice, • Small world Networks: By Watts & Strogatz

  4. Small World Graphs • Three main attributes used to analyze Small World Graphs : • Average Vertex Degree (k) (Avg. of No. of Edges Incident on ‘v’ over all ‘v’) • Average Characteristic Path Length (L) (Shortest Dist. B/w 2 points Avg. over all connected pairs) • Average Clustering Coefficient (C) (Prob. Of 2 nodes with a “mutual” friend being connected)

  5. Work of Watt and Strogatz • Asks why we see the small world pattern and what implications it has for the dynamical properties of social networks. • Their contribution is to show that the globally significant changes can result from locally insignificant network change.

  6. Watts -Strogatz (WS) Model (1998)

  7. Cohesive neighborhoods

  8. Mutual Clustering Coefficients • Cohesiveness or “cliquishness” of a graph • Originally, neighborhood cohesiveness around each vertex • In the paper, the neighborhood cohesiveness around individual edges

  9. Cvw • Cvw (mutual clustering coefficent) • For a pair of vertices v, w... • This coefficient is independent of the existence of an edge between v and w. • So, direct experimental evidence does not influence the assesment of neighborhood. • This measure is applied on edges, and on any pair of vertices.

  10. Cvw • Used for hypothesis about missing edges • 4 alternative definitions of Cvw are considered. • N(x) represents the neighborhood of a vertex x. • Total represents the total number of proteins in the organism.

  11. Cvw

  12. P value • The cumulative hypergeometric distribution is frequently used to measure • Cluster enrichment • Significance of co-occurence • The summation in the formula can be intrepreted as p value: • Tye probability of obtaining a number of mutual neighbors between vertices v and w, at or above the observed number by chance

  13. Protein-Protein interaction data • High-throughput, error-prone Y2H data • From CuraGen’s PathCallingYeast Interaction database • http://portal.curagen.com • For validation a more reliable conventional evidence used from PathCalling database. • Also Incyte Genomics’ Yeast Proteome Database is used for validation • http://www.incyte.com/proteome

  14. Cvw and validity

  15. Ranking by Cvw

  16. P+ • Compute the probability of an interaction being true, given the experimental evidence (Y2H) and local network topology (Cvw) • Estimate the probability that there is a high confidence evidence that the two proteins interact • It is likely an under-estimate

  17. P+ • This score can be computed by Bayes’ rule

  18. Predictions

  19. Pairs of proteins with high P+ score and no direct supporting evidence representr predicted interactions.

  20. Conclusion • Data containing errors • Local topology gives clues about confidence in networks • This approach is used to predict protein function • Can be generalized for other small world networks... • For finding the missing parts, or confidence levels..

  21. Thanx... • Questions ????

More Related