Humanmethylation450 hm450 arrays

We used data from TCGA collaboration as a source of gene expression (from RNA sequencing) and DNA methylation (from high-density methylation arrays). In order to obtain the gene expression profiles corresponding to each progression stage, we downloaded Illumina RNA-Seq level 3 gene expression files from TCGA–ccRC samples.
We used methylation data from TCGA in the form of beta values (β), which measure the level of DNA methylation at known CpG sites via Illumina HumanMethylation450 (HM450) arrays. These values are calculated from array intensities (level 2 data) as $\frac{M}{M+U}$ , where M corresponds to methylated probes. Meanwhile, U takes account for non-methylated ones, marked by bisulfite conversion (Zhou et al., 2016 (link)). The indexes of both datasets were harmonized to match patient codes as a key for paste RNA-seq and methylation beta values. This is the reason for the number of samples not corresponding with the original RNA-Seq number of samples. Download, annotation, and low-level analysis were performed using the TCGAbiolinks R library (Colaprico et al., 2015 (link)). We processed the clinical information directly from the TCGA-KIRC project. We categorized all samples by tumor_stage variable. The samples were cleaned to exclude those samples with non-reported stages or values. The TCGAbiolinks library was also used to retrieve clinical data from TCGA.

The gene expression profiles of 75 OSCC samples were obtained from The Cancer Genome Atlas (TCGA, www.cancergenome.nih.gov), including 65 OSCC tumor samples and 5 pairs of tumor and paracancerous samples. The HTSeq-Counts data were downloaded for subsequent analysis.
The DNA methylation data were downloaded from TCGA and Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/). The TCGA dataset consisted of the DNA methylation data and clinical information of 75 OSCC samples, including 65 OSCC tumor samples and 5 pairs of tumor and paracancerous samples. The GEO dataset GSE87053 included 11 OSCC tumor samples and 10 normal controls, and GSE123781 included 23 OSCC tumor samples and 18 normal controls. The DNA methylation profiles were all measured by Illumina Human Methylation 450 (HM450) arrays.