A Direct Clustering Method for Imperfect Microarray Data without Imputation

The existence of missing entries in microarray data is problematic for the proper clustering process. Several approaches have been introduced to overcome this problem. The main idea of those methods is the inclusion of imputation step during clustering analysis. However, these approaches are usually computationally expensive and badly imputed values can possibly mislead clustering results. In this work, we present a new clustering method which combines the separate clustering results of individual sample dimensions without the imputation of missing values. The performance of our method was superior to other typical clustering methods when it was tested with one model dataset and four microarray datasets.