Biclustering performs simultaneous clustering on features and data. It automatically integrates feature selection to clustering without any prior information, so that the relations of clusters of unsupervised labels (for example, genes) and clusters of data (for example, samples or conditions) are established. However, typical approaches have NP-complete computational complexity, which raises a great challenge to computational methods when identifying such local relations. This invention demonstrates that a neural-based classifier can be modified to perform biclustering in an efficient way. Experimental results on multiple human cancer data sets show that the algorithm can achieve clustering structures with higher qualities than those with other commonly used biclustering or clustering algorithms.

File Number: 11MST001 

Web site: http://ecodevo.mst.edu/

Other Information:

Case Manager: Keith Strassner (kdstrass@mst.edu)