Figure 4

Partitioning of ER expression tumors for breast cancer data. Tumor groupings after re-ordering for OREO, nsNMF after sorting of the first and second basis genes (as denoted by nsNMF F1 and nsNMF F2, respectively), hierarchical clustering after optimally re-ordering the leaves (HC (opt)), and hierarchical clustering (HC). In the data matrix, the white elements denote ER positive tumors, the black elements denote ER negative tumors, and the grey element denotes as tumor for which ER expression is unknown. All methods result in a moderately successful separation of ER positive and ER negative tumors, with OREO resulting in the richest grouping of ER negative tumors.