Ideker02 seems to combine K-means and simulated annealing for network clustering.
From Ma, 2007 Bioinformatics CGI paper:
Gene expression data and protein interaction data have been
integrated for gene function prediction. For example, Ideker
et al. (2002) used protein interaction data and gene expression
data to screen for differentially expressed subnetworks between
different conditions. In Tornow and Mewes (2003) and Segal
et al. (2003), gene expression data and protein interactions are
used to group genes into functional modules. These methods provide
insights into the regulatory modules of the whole networks at
the systems biology level. However, it is not clear how to adapt their
methods to identify genes contributing to the phenotype of interest.
Morrison et al. (2005) adapted the Google search engine to prioritize
genes for a phenotype by integrating gene expression profiles
and protein interaction data. However, the algorithm ignores the
information from proteins linked to the target protein through other
intermediate proteins, referred to in the rest of this paper as indirect
WGCNA: an R package for weighted correlation network analysis.