Cluster analysis of biological networks

Clustering methods are widely used as a first step for taming the complexity of biological networks and for extracting information from them. However, the bewildering number of available clustering algorithms and the lack of agreed upon validity measures represent a major challenge for the user, as no one-size-fits-all solution to the clustering problem exists. This article surveys the main types of clustering algorithms, highlighting their strengths and weaknesses when used for clustering biological data. We also discuss internal and external validity measures, emphasizing the importance of domain knowledge for choosing an appropriate clustering approach.