A dynamic approach for hierarchical clustering of gene expression data

Discovering patterns in gene expression data is an extremely important step in understanding functional genomics and it can be achieved through a clustering process. Biological processes are dynamic, therefore the data is continuously subject to change. Researchers can either wait until all data is available, or analyze it gradually, as the experiment progresses. Currently, the latter can only be accomplished by repeating the clustering process from the beginning. This would be very time consuming and could lead to important delays, considering the huge amounts of data to be dealt with. In this article we propose a dynamic approach for hierarchical clustering of gene expression data, which can handle the newly arrived data by adapting a previously obtained partition, without the need of re-running the algorithm from scratch. The experimental evaluation is performed on a real-life gene expression data set and the performance of our model is shown by the obtained results, which are analyzed in terms of several evaluation measures.