Fig. 3

Comparison between different cell isolation methods and published datasets. Box plots showing the distribution of coverage (A), mapping proportion (B), index of dispersion (C) among FACS-isolated, LCM-isolated and LCM-isolated blank samples. P-values were calculated using Kruskal–Wallis test among groups and Wilcoxon rank-sum test between groups. (D) Violin plot showing the distribution of coverage among different datasets. P-values were calculated using Wilcoxon rank-sum test between groups. This study, including only LCM: blue; van den Bos 2016: brown; McConnell 2013: purple. (E) A principal components analysis (PCA) was performed using the normalized read counts across autosomal bins (n = 5243) in published datasets and this study. Because they dominated the PCs, cells deviating from the [1.9–2] range were not included in the analyses. The number of cells for each dataset are indicated on the plot. X-axes illustrate PC1 and PC3 that explain 18.4% and 1.3% of the total variance, respectively. Y-axes show PC2 and PC4 that explain 2.8% and 0.9% of the total variance, respectively. (F) Boxplots showing the distribution of median CN of chromosome 1 (chr1, upper part of the figure) and chromosome 21 (chr21, lower part of the figure) across bins (n = 440 and n = 68 for chr1 and chr21, respectively). Each point corresponds to the median CN of each cell. Minimum (“Min”), median (“Med”), maximum (“Max”) and standard deviation (“sd”) of each distribution were shown on the boxplot. Cells that deviated from the [1.9–2] range were excluded from the analyses to be consistent with our filtering criteria (except for the uncorrected datasets). This study [Uncorrected (n = 1337), Uncorrected-filtered (n = 588), PCA-corrected (n = 1301)]: blue; van den Bos 2016 (n = 1468): brown; McConnell 2013 (n = 109): purple