Fig. 3

Deconvolution of TCGA-AML and healthy samples using LinDeconSeq. a Expression of the signature matrix. The expression was row-normalized. The upper bound of the color bar is 1. b-c t-SNE for all genes (b) and t-SNE for the signature genes (c), respectively. Each scatter represents a FACS-purified cell sample. d Circular bar plot of the cellular fractions for the 179 TCGA-AML patients. Each bar represents a sample and each color represents a specific cell type. The meanings of the colors are same as that in the legend of Fig. 2a. e The cellular fractions predicted by LinDeconSeq and CIBERSORT on the TCGA-AML patients. Each point represents a specific cell type in the sample. Pearson correlation coefficient (PCC, r) was calculated between the cellular fractions by LinDeconSeq and CIBERSORT. f The fractions of 13 primary blood cell types in AML and healthy samples. Within each group, each scatter represents the fraction of a specific cell type. The thick line in the box represents the median value. The bottom and top of the boxes are the 25th and 75th percentiles (interquartile range). The whiskers encompass 1.5 times the interquartile range. The statistical difference of the two groups was compared through the Wilcox test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. g ROC curve measuring the predictive based on the different cellular fractions between AML and healthy samples. The areas under (AUC) of ROC curve are 0.911, 0.985 and 0.984 for SVM, Random Forest and Logistic Regression, respectively