Dataset Information

Bulk RNA-Seq gene expression profiles from 13 TCGA cancer types and normal tissues for tumor classification

ABSTRACT: This dataset includes bulk RNA-Seq gene expression profiles of 6310 samples from 13 cancer types and normal tissues, obtained from The Cancer Genome Atlas (TCGA) and processed using a standardized pipeline. The raw count matrix, initially containing 60,660 genes, was filtered to retain only valid ENTREZ gene identifiers matched via the org.Hs.eg.db annotation package. Genes with zero variance across samples were removed. Normalization was performed by transforming expression values into log2 Counts Per Million (CPM) using the edgeR::cpm() function with log transformation. To ensure all values were positive and suitable for downstream modeling, a global shift was applied by adding the absolute minimum value. Subsequently, Gene Ontology (GO) analysis was performed across all three categories (Biological Process, Molecular Function, Cellular Component) using the TopGO package. Significant genes were selected, followed by Benjamini–Hochberg p-value correction, and validated using both Fisher’s and Kolmogorov–Smirnov tests. The resulting expression matrix consists of 18,564 functionally relevant genes across 6310 samples and forms the basis for clustering, classification, and metric evaluation within hybrid modeling frameworks for cancer diagnostics.

ORGANISM(S): Homo sapiens

PROVIDER: GSE304485 | GEO | 2026/02/12

REPOSITORIES: GEO

ACCESS DATA

Dataset's files

Source:

			Action	DRS
		Other

Items per page:

1 - 1 of 1

Similar Datasets

Project description:Samples from 20 children with B-cell precursor ALL, comprising ten with high hyperdiploidy and ten with the ETV6/RUNX1 fusion, were included in the study. Four Âµg of methylated and input fraction DNA were analyzed with a classic MeDIP assay and subsequent microarray hybridization using the NimbleGen CpG Island-Plus-Promoter Array (Roche NimbleGen, Madison, WI).<br><br>This array platform covers all 28,226 UCSC-annotated CpG islands and promoter regions for all RefSeq genes, including 385,000 probes of 50-75mer length per array, with an upstream and downstream promoter tiling of 800 bp and 200 bp, respectively. Signal intensity data were extracted from the scanned images. Each feature on the array has a corresponding log2 ratio, which is the ratio of the input signals for the methylated DNA and the input fraction DNA. The log2 ratio was scaled in order to center the ratio data around zero by subtracting the bi-weight mean for the log2 ratio values for all features on the array from each log2 ratio value. A modified ACME algorithm, where a fixed-length window (750 bp) is placed around each consecutive probe, and the one-sided Kolmogorov-Smirnov test were then applied on the scaled log2 ratio data to determine whether the probes were drawn from a significantly more positive distribution (peak) of intensity log2 ratios than the other probes in the array. The resulting score for each probe is the -log10 p-value from the windowed Kolmogorov-Smirnov test around that probe. Peak data were generated by searching for probes above a p-value minimum cut-off (-log10) of 2.<br><br>Peaks within 500 bp of each other were merged. Each annotated gene was subsequently searched for peaks appearing in a specified promoter region around the transcription start site. The region searched was design-specific, spanning from 5 kb upstream to 1 kb downstream of the transcription start site. Genes that had at least two probes with peaks above the p-value minimum cut-off were considered to have significant CpG island methylation scores for hypermethylation.

Dataset Information

Bulk RNA-Seq gene expression profiles from 13 TCGA cancer types and normal tissues for tumor classification

Dataset's files

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets