Clustering of gene expression data and end-point measurements by simulated annealing

Most clustering techniques do not incorporate phenotypic data. Limited biological interpretation is garnered from the informal process of clustering biological samples and then labeling groups with the phenotypes of the samples. A more formal approach of clustering samples is presented. The method utilizes simulated annealing of the Modk-prototypes objective function. Separate weighting terms are used for microarray, clinical chemistry, and histopathology measurements to control the influence of each data domain on the clustering of the samples. The weights are adapted during the clustering process. A cluster's prototype is representative of the phenotype of the cluster members. Genes are extracted from phenotypic prototypes obtained from the livers of rats exposed to acetaminophen (an analgesic and antipyretic agent) that differed in the extent of centrilobular necrosis. Map kinase signaling and linoleic acid metabolism were significant biological processes influenced by the exposures of acetaminophen that manifested centrilobular necrosis.