Figure 2 from Noninvasive Lung Cancer Subtype Classification Using Tumor-Derived Signatures and cfDNA Methylome
Year: 2024
Type: other
Abstract: Identification and validation of the subtype-specific methylation markers. A, Heatmap of the logarithm-scaled normalized read count at the markers in the tumor tissue RRBS data. For lung adenocarcinoma and LUSC, we selected the top 2,500 hypermethylation and hypomethylation regions using the fragment-based marker discovery method. The color in the heatmap represents the logarithm-scaled normalized count of the hypermethylated (hypomethylated) reads at the hypermethylation (hypomethylation) markers. B, Heatmap of the logarithm-scaled normalized read count at the markers in the independent tumor tissue RRBS data. C, Heatmap of the methylation levels at the markers in TCGA methylation data. Methylation levels were quantified as the average β value across all probes covered by a marker. D, Heatmap of the logarithm-scaled transcription level at the marker-associated genes in the TCGA RNA-seq data. A gene is marker-associated if its promoter has more than 50% overlap with a marker. Every row in the heatmap represents a marker-associated gene, and every column represents a patient with cancer. A transcription level was quantified as the TPM. E, Differential gene expression between lung adenocarcinoma and LUSC in TCGA RNA-seq data. The statistical significance of the genes was determined by the Benjamini–Hochberg adjusted P values at 0.05. The adjusted P values are shown in a negative logarithm (base 10) scale. F, The discriminative power of the whole marker set was evaluated using a 10-fold cross-validation on tumor tissues. The discriminative power was quantified as the ROC curve, shown as the bold black curve. The gray area represents the CI of the ROC curve. The AUC was calculated, and the corresponding 95% DeLong CI is shown in the parentheses.
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: gold
OpenAlex ID: W4400700245