Fig. 2: Methylation versus cell type fraction in the Wang Discovery Set and FORECEE dataset. | Nature Communications

Fig. 2: Methylation versus cell type fraction in the Wang Discovery Set and FORECEE dataset.

From: Validation of blood-based detection of breast cancer highlights importance for cross-population validation

Fig. 2

a Cell type fractions inferred with EpiDISH in Breast Cancer (Wang, FORECEE; BC) and Systemic Sclerosis (GSE117929; SSc) cases and respective controls. Wang, n = 30 controls, n = 50 cases. FORECEE, n = 208 controls, n = 102 cases. GSE117929, n = 13 controls, 18 cases. b Methylation of 8 loci versus inferred granulocyte proportion. Differences in coverage between the 450k and EPIC arrays account for any missing data. c ROC curves of granulocyte proportion as a predictor of case/control status. Abbreviations: Bmem memory B cell, Bnv naive B cell, CD4Tmem memory CD4 T cell, CD4Tnv naive CD4 T cell, CD8Tmem memory CD8T cell, CD8Tnv naive CD8 T cell, Epi epithelial cell, Fib fibroblast, Granulocytes granulocytes (sum of neutrophils, eosinophils, and basophils), NK natural killer cell, Treg regulatory T cell. Boxplots are standard Tukey representation, whereby whiskers denote minima and maxima (smallest and largest values within 1.5 times the interquartile range), the box denotes the interquartile range (25th percentile, median and 75th percentile), and dots indicate outlier values (>1.5 times the interquartile range).

Back to article page