What it does

The program builds a network of gene categories connected by shared genes. The edges of this network are weighted based on the number of genes that each node shares. The clustering coefficient, c_u, is then calculated for each node using the formula:

where deg(u) is the degree of u and edge weights, w_uv, are normalized by the maximum weight in the network. The cluster coefficients are then filtered by our program based on threshold (that could be a percentile or a value choose by the user) and all the nodes with a cluster coefficient lower than this threshold are deleted from the network. Finally, the program reports each connected component as a cluster of gene classifications. With our program a lower number of gene categories is obtained, but the results are easier to interpret as they exclude genes present in many gene groups.