Benchmarking a new semantic similarity measure using fuzzy clustering and reference sets: Application to cancer expression data

摘要: Clustering algorithms rely on a similarity or distance measure that directs the grouping of similar objects into same cluster and separation distant between distinct clusters. Our recently described semantic (IntelliGO), applies to functional comparison genes, is tested here for first time in clustering experiments. The dataset composed genes contained benchmarking collection reference sets. Heatmap visualization hierarchical illustrates advantages using IntelliGO over three other measures. Because often belong more than one clustering, fuzzy C-means also applied dataset. choice optimal number clusters performance are evaluated by F-score method Overlap analysis proposed as exploiting matching Finally, our list found dysregulated cancer samples. In this case, sets provided expression profiles. these profiles obtained with leads characterize subsets displaying consistent function