CREAMMIST

CREAMMIST integrates dose-response data from CCLE, GDSC1, GDSC2, CTRP1, and CTRP2 using a Bayesian framework to produce unified dose-response curves and quantify uncertainty for cancer drug sensitivity analysis in integrative pharmacogenomics.

Key Features:

• Integrative Dose-Response Curves: Produces unified dose-response curves across CCLE, GDSC1, GDSC2, CTRP1, and CTRP2 for comparative drug sensitivity assessment in cancer cell lines.
• Uncertainty Quantification: Quantifies uncertainty and high-certainty alignment across datasets via a Bayesian framework, capturing variability from experimental setups and preprocessing.
• Statistical Outputs: Generates statistical outputs derived from integrated dose-response curves for biomarker identification, drug concentration selection for experiments, and training robust machine learning models.
• Enhanced Data Utilization: Integrates over disparate datasets (over 14 million dose-response values) to maximize use of available pharmacogenomic data.

Scientific Applications:

• Predictive modeling of drug sensitivity: Supports development and training of models to predict cancer cell-line responses to drugs using integrated dose-response information.
• Biomarker identification: Provides statistical evidence from integrated curves to identify molecular biomarkers associated with drug response.
• Experimental design optimization: Informs selection of drug concentrations and experiment planning by consolidating dose-response evidence across datasets.

Methodology:

Implements a Bayesian framework to integrate over 14 million dose-response values from CCLE, GDSC1, GDSC2, CTRP1, and CTRP2, producing integrated dose-response curves and quantified uncertainty.