Clusters in protein interaction networks can potentially help identify functional relationships
among proteins. We study the clustering problem by modeling it as graph cut problems.
Given an edge weighted graph, the goal is to partition the graph into a prescribed
number of subsets obeying some capacity constraints, so as to maximize the total weight
of the edges that are within a subset. Identification of a dense subset might shed some light
on the biological function of all the proteins in the subset.
We study integer programming formulations and exhibit large integrality gaps for various
formulations. This is indicative of the difficulty in obtaining constant factor approximation
algorithms using the primal-dual schema. We propose three approximation algorithms for
the problem. We evaluate the algorithms on the database of interacting proteins and on
randomly generated graphs. Our experiments show that the algorithms are fast and have
good performance ratio in practice. / xiii, 71 leaves : ill. ; 29 cm.
| Identifer | oai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:ALU.w.uleth.ca/dspace#10133/774 |
| Date | January 2008 |
| Creators | Choudhury, Salimur Rashid, University of Lethbridge. Faculty of Arts and Science |
| Contributors | Gaur, Daya |
| Publisher | Lethbridge, Alta. : University of Lethbridge, Deptartment of Mathematics and Computer Science, 2008, Arts and Science, Mathematics and Computer Science |
| Source Sets | Library and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada |
| Language | en_US |
| Detected Language | English |
| Type | Thesis |
| Relation | Thesis (University of Lethbridge. Faculty of Arts and Science) |
Page generated in 0.0016 seconds