Raw sequencing data from selected tissue are processed and aligned to generate a scRNA-seq expression count matrix. Unsupervised clustering separates malignant and normal cell clusters using an ensemble prediction approach with three analytical tools: ScType, CopyKAT and SCEVAN (see Suppl. Figure 3 ). InferCNV infers large-scale copy number variations and identifies genetically distinct subclones among the malignant cells.