CTR-DB

CTR-DB catalogs 83 patient-derived pre-treatment transcriptomic datasets linking drug responses across 28 histological cancer types and 123 drugs (5,139 patient samples) to support analysis of molecular determinants of cancer treatment sensitivity and resistance.

Key Features:

• Extensive Data Collection: Aggregates 83 patient-derived pre-treatment transcriptomic datasets covering 28 histological cancer types, 123 drugs, and 5,139 patient samples.
• Manual curation: Drug response annotations and dataset metadata are manually curated to ensure high-quality response labels.
• Differential gene expression analysis: Performs differential expression analysis on individual datasets to identify genes associated with response or resistance.
• Receiver operating characteristic (ROC) curve generation: Calculates ROC curves to evaluate predictive performance of genes or signatures.
• Functional enrichment analysis: Conducts enrichment analyses to interpret differentially expressed genes in pathways and gene sets.
• Sensitizing drug search: Identifies candidate sensitizing drugs based on transcriptomic response signatures.
• Tumor microenvironment analyses: Analyzes tumor microenvironment features linked to drug response.
• Multiple-dataset integration and comparison: Supports combining and comparing multiple datasets to identify cross-study patterns and correlations.
• Biomarker validation: Provides functions to validate predictive biomarkers across datasets.

Scientific Applications:

• Molecular determinant analysis: Identifies genes and pathways associated with treatment sensitivity or resistance in cancer patients.
• Predictive biomarker discovery and validation: Supports discovery and cross-dataset validation of biomarkers predictive of drug response.
• Resistance mechanism heterogeneity studies: Enables comparative analysis of heterogeneous resistance mechanisms across cancer types and therapies.
• Therapeutic strategy and combination identification: Facilitates identification of potential therapeutic strategies and sensitizing drug combinations informed by transcriptomic signatures.

Methodology:

Computational methods explicitly include differential gene expression analysis, receiver operating characteristic (ROC) curve generation, functional enrichment analysis, sensitizing drug search, tumor microenvironment analyses, dataset integration and comparison, and biomarker validation.