clusterMaker2 is the Cytoscape 3 version of the clusterMaker plugin. clusterMaker2 provides several clustering algorithms for clustering data within columns as well as clustering nodes within a network. This version also provides support for two new algorithms: Fuzzy C-Means and a new "Fuzzifier". In addition to providing clustering algorithms, clusterMaker2 provides heatmap visualization of both node data and edge data as well as the ability to create new networks based on the results of a clustering algorithm.
Current node attribute algorithms:
* Hierarchical
* K-Means
* K-Medoid
* AutoSOME
* Create correlation network
* PAM
* HOPACH
* DBSCAN
The current network partitioning algorithms are:
* Affinity propagation
* Cluster "Fuzzifier"
* Connected Components
* Community Clustering (GLay)
* Fuzzy C-Means Clustering
* MCODE
* MCL
* SCPS (Spectral Clustering of Protein Sequences)
* Transitivity Clustering
four post-cluster filters:
* Best Neighbor filter
* Cutting Edge filter
* Density filter
* Haircut filter
five cluster ranking algorithms:
* Additive sum
* Multiply sum
* Page Rank
* Page Rank with priors
* Hyperlink induced topic search
and 3 dimensionality reduction techniques:
* PCA
* PCoA
* t-SNE
In addition, clusterMaker2 provides 5 ways to visualize the data:
* Create New Network from Attribute
* Create new Network from clusters
* JTree TreeView
* JTree KnnView
* JTree HeatMap view (unclustered)
clusterMaker2 1.1 improves performance and adds new algorithms as well as entirely new classes of analyses that allow users to rank clusters using orthogonal data and perform dimensionality reduction techniques, including PCA, PCoA, and t-SNE. We've also significantly improved the performance of some of the existing algorithms, most notably MCL.
This version of clusterMaker has several bug fixes, but more importantly it adds a new ResultsPanel visualization that shows the individual clusters for network partitioning algorithms. There are also some new algorithms, including the Cheng&Church bicluster algorithm.
