Bioinformatics data processing is a complicated process. This paper presents a data preprocessing step that analyzes the inner structures of a data set and that is called class decomposition. This step uses clustering (in this case applying hierarchical agglomerative clustering) to find high density areas in the classes present in data and re-labels them as subclasses. To implement class decomposition, the data is first analyzed using clustering then the clusters are evaluated and selected as subclasses. This article focuses on cluster stability evaluation to assess the characteristics of the data set and the found subclasses. The evaluation is an iterative process, making small changes to the data set in every step and reapplying cluster analysis. These small changes (removing one object from the data set repeated for 20 iterations in this case) should not have any impact on clusters if they are stable (meaning that other objects that were not removed stay in the same clusters as in the full clustering).