Tumor classification by gene expression profiling

Abstract

The considerable "algorithmic complexity" of biological systems requires a huge amount of detailed information for their complete description. Current high-throughput technology such as microarrays is generating an overwhelming amount of data of biological systems at the molecular and cellular level. To adequately organize, maintain, analyze and interpret this deluge of information the adaptation of existing and the development of new computational methodologies and tools is required. The principal approach to analyzing and interpreting biological data is to abstract them into logical structures that support and incrementally promote the development of a more general conceptual framework for characterizing, explaining, and predicting processes in living systems. Cluster analysis refers to a computing methodology that discovers and describes meaningful patterns or structures in data. Generally, cluster algorithms are governed by a learning-by-observation process. A plethora of specific algorithms has been suggested in the literature. In the context of microarray gene expression profiling of tumors, this work describes a comparative study of five clustering methods.