GSE45827: Breast Cancer Gene Expression with 6 Subtypes, 151 Samples

54676 gene expression measurements from 151 breast cancer tissue samples, curated into 6 cancer subtype classes. The data originates from the CuMiDa repository, which handpicked and preprocessed this dataset from the GEO database for machine learning benchmarking. CuMiDa's curation process involved steps like sample quality control, background correction, and normalization to create a reliable source.

Use Cases

• Benchmarking classification algorithms based on the 6 cancer subtype labels provided.
• Identifying important genes for cancer subtype classification based on the 54676 gene expression features.
• Exploring potential novel cancer subtypes through clustering analysis of the gene expression profiles.
• Addressing class imbalance challenges in multi-class classification using the provided sample distribution.

Strengths

• Includes 151 samples with 54676 gene expression features, providing high-dimensional data for analysis.
• Manually curated from the GEO database with steps for sample quality, background correction, and normalization.
• Part of a repository (CuMiDa) that provides benchmark results from 3-fold cross-validation for comparison.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• The original sample collection methodology and potential demographic or geographic biases are not described.

Provenance

Source

Gene Expression Omnibus (GEO) accession GSE45827, curated by the CuMiDa repository.

Collection Method

Handpicked from approximately 30,000 GEO studies, with biological preprocessing applied.