Zeckzer, D., Wiegreffe, D., Müller, L.
Abstract:
A major goal in epigenetics is understanding how cells differentiate into different cell types. Besides the increase
of individual data sets, the amount of replicated experiments generating a tremendous amount of data is ever
increasing. While biologists primarily analyze their data on the highest level using statistical correlations or on the
lowest level analyzing nucleotide sequences, determining the fate of histone modifications during cell specification
necessitates improved analysis capabilities on one or more intermediate levels. For this type of analysis, it proved to
be very useful to use tiled binned scatter plot matrices showing binary relationships or to use tiled binned 3D scatter
plots showing ternary relationships. Quarternary or general n-ary relationships are not easily analyzable using
visualization techniques like scatter plots, only. Therefore, we augmented existing clustering methods with the
tiling and binning idea enabling the analysis of n-ary relationships. Analyzing the changes of histone modifications
comparing two cell lines using tiled binned clustering, we found new, unknown relations in the data.