Clustering / SOM


Self-organizing maps (SOM) are a method were data dimensions are reduced using self-organizing neural networks. The current implementation uses Euclidean distances between genes in assigning them to different cells of the network.



Note that individual cells do not necessarily represent individual clusters. Rather, cells that are colored with a similar color probebly from a coherent cluster.

In the som.png image Property visualizes the internode distances. Codes shows the gene expression profiles. Counts gives information on how many genes there are in each node (red being the smallest value, anf yellow the highest). Mapping shows were genes are mapped.


A file describing the SOM grid that can be visualized using Chipster GUI. In addition, clustering result as rendered with R is presented.