Normalized lmQCM: An Algorithm for Detecting Weak Quasi-Cliques in Weighted Graph with Applications in Gene Co-Expression Module Discovery in Cancers.

ABSTRACT: In this paper, we present a new approach for mining weighted networks to identify densely connected modules such as quasi-cliques. Quasi-cliques are densely connected subnetworks in a network. Detecting quasi-cliques is an important topic in data mining, with applications such as social network study and biomedicine. Our approach has two major improvements upon previous work. The first is the use of local maximum edges to initialize the search in order to avoid excessive overlaps among the modules, thereby greatly reducing the computing time. The second is the inclusion of a weight normalization procedure to enable discovery of "subtle" modules with more balanced sizes. We carried out careful tests on multiple parameters and settings using two large cancer datasets. This approach allowed u

SUBMITTER: Zhang J 

PROVIDER: S-EPMC4962959 | biostudies-literature | 2014

REPOSITORIES: biostudies-literature