Phylogenetic analysis of cancer microarray data
Abstract
Recent advances in biotechnology and the availability of genetic data have greatly facilitated the molecular exploration of cancer. Cancer microarray data analysis provides new insights into the correlation between gene expression changes with normal to cancerous cells conversion or with cancer treatment outcomes.
In this study, we have applied phylogenetic methods for cancer classification using microarray data. We compute a reduced distance matrix from the phylogenetic tree inferred by maximum likelihood estimation, and apply existing clustering methods to classify samples. In addition, we develop a new classification method based on the estimated phylogenetic tree. Our method performs equally well for the classification problem with much fewer number of model parameters, compared to the classification methods based on the Pearson correlation matrix.