Validation_real: Four RNA-seq Datasets with Ground Truth Cell-Type Fractions

Four RNA-seq datasets from human and mouse tissues, accompanied by ground truth cell-type fractions. The collection is 131.7 MB in size, provided by Alexander Dietrich and last updated on June 3, 2026. Gene expression values are available as either raw counts or TPM normalized counts.

Use Cases

• Benchmarking cell-type deconvolution tools based on the provided ground truth fractions.
• Training machine learning models for cell-type prediction using the raw and TPM normalized gene expression counts.
• Comparing gene expression profiles across different organisms and tissues, such as human blood and mouse spleen.
• Validating the performance of new normalization methods against established ground truth data.

Strengths

• Includes ground truth cell-type fractions for validation, a key feature mentioned in the description.
• Provides data in two common formats for gene expression analysis: raw counts and TPM normalized counts.
• Aggregates data from multiple published studies, with nine distinct source datasets listed.
• Dataset size is 131.7 MB, indicating a medium-scale collection suitable for local analysis.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for specific modeling tasks.
• Data may reflect bias inherent to the specific source studies and experimental methods (e.g., FACS, CBC).

Provenance

Source

Aggregated from nine published studies on figshare, with DOI links provided.

Collection Method

Likely compiled from public RNA-seq experiments where cell-type fractions were measured via techniques like FACS, CBC, or population purification.

Freshness

Last updated 2026-06-03 09:53:16; freshness should be verified.