FGFR Inhibitor Sensitivity in Cholangiocarcinoma with Interpretable Machine Learning

A pharmacogenomic dataset integrating large-scale drug sensitivity screening with baseline transcriptomes for cholangiocarcinoma. The data includes a composite FGFR inhibitor sensitivity score derived from multiple compounds and an interpretable transcriptional program extracted from linear model coefficients. The dataset was created by Yading Xie and last updated on 2026-04-30.

Use Cases

• Training predictive models for FGFR inhibitor sensitivity based on baseline transcriptomic profiles.
• Validating pharmacogenomic biomarkers across different platforms like PRISM and GDSC.
• Extracting interpretable gene signatures associated with drug response pathways.
• Assessing the reliability of composite drug sensitivity scores versus single-drug readouts.

Strengths

• Dataset employs a composite FGFR inhibitor score showing strong split-half reliability (median ρ = 0.63).
• Modeling used strict group-aware cross-validation to prevent information leakage.
• Includes cross-platform validation using an orthogonal resource (GDSC) with matched cell lines.
• A reduced 15-feature panel preserved concordance with the full signature in external cohorts.

Limitations

• Row count and specific column definitions are unknown, requiring manual inspection after download.
• The dataset is very small at 17.1 KB, indicating limited scope or summary-level data.
• Description metadata is detailed but lacks column-level documentation for field semantics.

Provenance

Source

figshare, authored by Yading Xie.

Collection Method

Integrated large-scale pharmacogenomic screening with baseline transcriptomes; predictors trained using strict group-aware cross-validation.

Freshness

Last updated 2026-04-30 05:24:00; freshness should be verified.