Hierarchical clusters of liver and non-liver cancers based on distinguishing CNV-features. (A) Clustering using localized CNV-features and (B) Clustering using recurrent CNV-features. The 21 liver and 16 non-liver cancers analyzed are described in Additional file 2: Table S2. The distinguishing localized and recurrent CNV-features selected by machine learning for the purpose of clustering these two classes of cancers are listed in Additional file 6: Table S3A and 3B respectively. The numbers in orange shown at the nodes for the ‘liver cancer’ (blue solid box) and ‘non-liver cancer’ (green dashed box) clusters indicate the approximate unbiased probabilities, and the three incorrectly clustered samples in Part (B) are shown in red. Clustering of samples was performed as described in Methods.