Improvement of trustworthiness of the yeast data visualizations when q least trustworthy genes are discarded from the visualization. Initially, the size of the neighborhood was k = 20 but was gradually decreased to keep its ratio to the number of the remaining data points constant. Sammon: Sammon's mapping, NMDS: non-metric multidimensional scaling, SOM: self-organizing map, HC: hierarchical clustering, with the ultrametric distance measure and with the linear distance measure. The sudden steps in the trustworthiness of hierarchical clustering coincide with the changes of the neighborhood radius k.
Kaski et al. BMC Bioinformatics 2003 4:48 doi:10.1186/1471-2105-4-48