Comparison of Phylogenetic and Phenetic Clustering
Mentor: Dr. Terry Hyslop, Department of Biostatistics and Bioinformatics, Duke University
Date/Time: August 27, 2019 at 2pm
Location: Room 1300, Harris Building
In expression analysis, Phylogenetics is used to infer evolutionary relationships between samples. Phylogenetic cladograms clusters samples into clades based on shared derived states. States are based on expression values that are either within or outside of a normal (healthy expression) range. This type of clustering allowed for a directional classification of 15 breast cancer (BC), reduction mammoplasty (RM), and prophylactic mastectomy (PM) HU113A Affymetrix expression samples and 2 progressor samples from a clinical study run on both Nanostring and Affymetrix Clarium S arrays. Euclidean based hierarchical clustering (HC) revealed comparable clustering. Another hierarchical dendrogram technique, using the hclust function in R provided similar clustering without gene expression information. Lastly, GenePattern expression heatmap provided clustering of genes based on user provided classes. Whereas, the phlylogenetic cladogram provided directional classification without data sub setting based on predetermined classes and without statistical manipulation of raw data. The analysis of different array was unreliable due to batch effect which normalization and scaling could be used to offset in future analysis.