Figure 1 from Noninvasive Lung Cancer Subtype Classification Using Tumor-Derived Signatures and cfDNA Methylome

Overview of the data usage and study design. A tissue RRBS dataset (n = 60) was used to identify subtype-specific methylation markers. We filtered the markers that were not differential between the solid tumor tissues and the noncancer plasma samples (n = 30). The remaining subtype-specific markers were then validated using an independent tissue RRBS dataset (n = 43), TCGA project’s tumor tissue methylation data (n = 794), and RNA-seq data (n = 1,029). We applied the subtype-specific markers identified from tumor tissue to train and validate the subtype classification model on plasma cfDNA using the cfMethyl-Seq data of the patients with lung cancer (n = 106). Then we validated this cfDNA-based classification model on an independent cfMethyl-Seq dataset (n = 27).