Figure 3From: mCOPA: analysis of heterogeneous features in cancer expression dataAnalysis of the number of samples sharing a given outlier. Most outlier features are outliers in only a small number of metastatic samples, with very few outliers shared across more than three samples. Very similar proportions are observed when sample counts for either under- or over-expressed outliers are considered separately. Counts are only shown for those outliers that occur in the metastasis cluster. A further 201 outliers map exclusively to non-metastatic samples.Back to article page