%0 Journal Article %T WebGimm: An integrated web-based platform for cluster analysis, functional analysis, and interactive visualization of results %A Vineet K Joshi %A Johannes M Freudenberg %A Zhen Hu %A Mario Medvedovic %J Source Code for Biology and Medicine %D 2011 %I BioMed Central %R 10.1186/1751-0473-6-3 %X Identifying groups of co-expressed genes through cluster analysis has been successfully used to elucidate affected biological pathways and postulate transcriptional regulatory mechanisms. Methods for co-expression analysis of gene expression data have been extensively researched, and numerous clustering algorithms have been developed. New clustering algorithms often have been implemented as stand-alone computer programs, R packages, or both [1]. Numerous open source and commercial integrated analysis systems also implement multiple clustering algorithms. For example, MultiExperiment Viewer (MeV) [2] provides access to several clustering procedures as well as the mechanism for adding additional methods. The MeV+R package expands the utility of MeV to serve as a general "wrapper" and GUI for Bioconductor R packages [3]. Several web-servers for using specific clustering procedures exist where the web-interface is designed to gather data and necessary parameter values while the actual computation is performed on remote servers [4,5]. Separating the user interface from the computational infrastructure executing the algorithm, allows for computationally efficient implementations that utilize high-end HPC infrastructure to be leveraged against often computationally demanding clustering algorithms. Despite all these efforts, the methods most commonly used in practice are simple hierarchical clustering procedures implemented in Michael Eisen's cluster programs [6]. Results typically are visualized using the associated treeview program. "Interesting" clusters are selected by visual inspection, and functional enrichment analysis, if any, is performed using well-established online resources such as DAVID [7]. While seemingly ad-hoc, such general strategy has been remarkably successful in the analysis of genomics data.The rationale for developing WebGimm is two-fold. First, sophisticated and better performing clustering methods are likely to be used more often if they are access %U http://www.scfbm.org/content/6/1/3