Figure 3.

The distribution of feature-weighted similarities between ovarian cancer cell lines and ovarian (HGSOC) tumors

(A–D) (A) The overall similarity represents the average of the feature-weighted similarity over the three datum types: (B) mRNA expression (with higher weights on genes with the largest expression changes); (C) mutations; and (D) copy-number alterations (with higher weights on the most recurrently altered genes in (C and D)). The distributions of feature-weighted similarities using mutations reveal striking differences between cell lines, with low, high, and intermediate similarities to tumors for three groups of cell lines. Cell lines, such as OAW-28 and Kuramochi have a high overall similarity to tumors ranking as they rank highly across all three datum types, whereas cell lines, such as PA-1 and TYK-nu, are poor matches to tumors across multiple datum types in terms of genomic similarity over the genes used in the feature-weighted similarity measure.