GSE50161: Brain Cancer Gene Expression with 5 Subtypes, 130 Samples

A gene expression dataset for brain cancer containing 130 samples and measurements for 54,676 genes, curated into 5 classes. It originates from the CuMiDa repository, which provides handpicked and biologically preprocessed microarray datasets from the Gene Expression Omnibus (GEO) for machine learning. The dataset is referenced in computational biology publications from 2019.

Use Cases

• Benchmarking classification algorithms based on the 5 cancer subtypes.
• Identifying key biomarker genes for brain cancer classification based on the 54,676 gene features.
• Exploring cancer subtype discovery through clustering techniques on the gene expression profiles.
• Addressing class imbalance challenges in multi-class medical datasets.

Strengths

• Contains 54,676 gene expression features, providing a high-dimensional feature space for analysis.
• Includes 130 samples across 5 distinct cancer classes.
• Manually curated from over 30,000 GEO studies with standardized biological preprocessing for reliability.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for some modeling tasks.
• Last update date is unknown; freshness unverified.

Provenance

Source

Gene Expression Omnibus (GEO) accession GSE50161, curated by the CuMiDa (Curated Microarray Database) project.

Collection Method

Handpicked from GEO and subjected to biological preprocessing including sample quality control, background correction, and normalization.