Inference of copy number variations and clonality analysis.

To obtain the inferred CNV result suitable for clonality analysis, the "subcluster" mode in inferCNV was utilized. Please refer to the attached "inferCNV.r" file for some example script to run inferCNV. A 6-state CNV model was adapted by inferCNV, and each detected CNV was predicted to be in one of the following CNV levels: State 1, complete loss; State 2, loss of 1 copy; State 3, neutral; State 4, addition of 1 copy; State 5, addition of 2 copies; and State 6 addition of more than 2 copies. The CNV was considered as “loss” for States 1 and 2 or “gain” for States 4, 5, and 6 events. Each CNV events detected was summarized in largescale genomic regions and attributed to the chromosome arm levels according to the hg38 cytoband information (http://hgdownload.cse.ucsc.edu/goldenpath/hg38/database/). The identified CNV types and calculated percentage of cells within each CNV type were processed in R and visualized using Uphyloplot2 (https://github.com/harbourlab/UPhyloplot2). Specifically, files (HMM_CNV_predictions.HMMi6.rand_trees.hmm_mode-subclusters.Pnorm_0.5.pred_cnv_regions and 12_HMM_preds.cell_groupings) were needed for the data wrangling in R. Finally, the cell grouping information from UPhyloplot2 (cell percentage in each subgroup) was processed and summarized in R to generate the table as shown in Supplementary Table 4 in the original publication. For the readers' convenience, I am posting some example script that I used in R for the downstream analysis using the data from inferCNV and UPhyloplot2.