Reducing the Subjectivity of Gene Expression Data Clustering Based on Spatial Contiguity Analysis

* Final gross prices may vary according to local VAT.

Get Access


Clustering, which has been widely used as a forecasting tool for gene expression data, remains problematic at a very deep level: different initial points of clustering lead to different processes of convergence. However, the setting of initial points is mainly dependent on the judgments of experimenters. This subjectivity brings problems, including local minima and an extra computing consumption when bad initial points are selected. Hence, spatial contiguity analysis has been implemented to reduce the subjectivity of clustering. Data points near the cluster centroids are selected as initial points in this paper. This accelerates the process of convergence, and avoids the local minima. The proposed approach has been validated on benchmark datasets, and satisfactory results have been obtained.