Refinement of Subnetwork Discovery Algorithm for Biological Networks

The abundance of biological experimental data from new high-throughput technologies, such as microarray data, suggests the need for new methods to study biological processes using a systems-based approach. Handling this huge volume of data requires the development of new computational methods to analyze and extract interesting pieces of biological information. JAM (jActiveModules) is a Cytoscape plugin developed to nd connected sets of genes with high levels of di erential expression. This network approach helps biologists to generate new hypotheses concerning the biological mechanisms underlying observed changes in gene expression. In this work, the search algorithm of JAM is modi ed and measured. The goal was to improve the sensitivity and speci city of the method compared to the original JAM algorithm. The modi cations made to the search algorithm involve: randomizing the starting point of the search, constraining the number of subnetworks maintained while searching, and improving the regional-scoring heuristic. Importantly, these modi cations increase the number of signi cant genes observed in the results. To ensure consistency in the search results, we apply the search algorithm multiple times and develop a statistical lter to retain consistent genes appearing across JAM runs. Furthermore, we apply this improved version of JAM to DC (Dendritic Cell) maturation microarray data and KEGG pathways to study the underlying mechanisms behind the DC maturation process, an essential part of the development of protective immunity to a number of infectious pathogens.

Grantor dc:publisher

Wake Forest University

Year dc:date.issued

2010

Author dc:creator

• Sajuthi, Satria

• Computer Science

Language dc:language.iso

en_US

Handle dc:identifier.uri

http://hdl.handle.net/10339/14884

OAI identifier oai:identifier

oai:wakespace.lib.wfu.edu:10339/14884