TG-LASSO

TG-LASSO predicts clinical drug response in cancer patients by integrating tissue-of-origin information with gene expression profiles using LASSO regularized regression to provide tissue-guided prediction and biomarker discovery.

Key Features:

• Integration of Tissue Origin: Incorporates tissue-of-origin information alongside gene expression profiles to provide tissue-specific context for drug-response prediction.
• Regularized Regression Approach: Employs LASSO regularized regression to limit overfitting and manage multicollinearity in high-dimensional gene expression data.
• Preclinical Sample Utilization: Trains models on preclinical in-vitro cancer cell line (CCL) data to enable prediction when large-scale patient datasets are limited.
• Performance and Validation: Systematically evaluated against various linear and non-linear algorithms using extensive databases, improving discrimination between drug-resistant and sensitive patients.
• Biomarker Identification: Identifies genes associated with drug response, including known drug targets and relevant pathways.
• Clinical Relevance: Links genes associated with tissue-specific drug responses to patient survival outcomes.

Scientific Applications:

• Predict Drug Efficacy: Forecasts patient response to anticancer drugs from molecular profiles to inform treatment selection.
• Identify Biomarkers: Pinpoints genes linked to drug sensitivity and resistance for potential diagnostic or therapeutic biomarker development.
• Enhance Preclinical Studies: Validates and refines preclinical in-vitro models by comparing CCL-derived predictions to clinical outcomes.

Methodology:

Integrates tissue-of-origin information with gene expression profiles and applies LASSO regularized regression, training on preclinical in-vitro cancer cell line (CCL) data and evaluating models against linear and non-linear algorithms using extensive databases.