two methods (affine propagation and k-means)
The function compute clusters based on sequence similarity defined by the feature vector. Two unsupervised clustering algorithms are used, affine propagation and k-means. The number of clusters for k-means is set to be the same as for affine propagation. Number of clusters for affine propagation is controled by the config.ap_clustering_factor variable. The config.ap_clustering_factor should be set such that small pertubation of this variable does not change the number of clusters (significantly).
Input:
config (structure): config structure
sequence (structure): A valid sequence structure.
similarity (matrix): similarity between every pair of the sequence feature vector
feature_vectors (matrix): feature vector for each sequence
Output:
clusters_ap (structure): clustering of the affine propagation method
clusters_kmeans (structure): clustering of the k-mean method
Clusters structure:
cluster_map (vector): assignment of each sequence to some exemplar
exemplars (vector): representants of the clusters
clusters_id (cell vector): ids of sequences for each cluster