Clustering gene expression data with the PKNNG metric

In this work we use the recently introduced PKNNG metric, associated with a simple Hierarchical Clustering (HC) method, to find accurate an stable solution for the clustering of gene expression datasets. On real world problem it is important to evaluate the quality of the clustering proccess. According to this, we use a suitable framework to analyze the stability of the clustering solution obtained by HC + PKNNG. Using an artificial problem and two gene expression datasets, we show that the PKNNG metric gives better solutions than the Euclidean method and that those solutions are stable. Our results show the potential of the association of the PKNNG metric based clustering with the stability analysis for the class discovery process in high throughput data Workshop de Agentes y Sistemas Inteligentes (WASI) Red de Universidades con Carreras en Informática (RedUNCI)