Precision oncology: Computational methods for multi-omics data integration to improve drug response prediction

Area of Science:

• Computational biology
• Genomics
• Precision medicine

Background:

• Cancer heterogeneity complicates drug treatment, necessitating personalized approaches.
• Multi-omics data integration advances precision medicine by enhancing tumor biology understanding.
• Integrating diverse omics data presents computational challenges for traditional methods.

Purpose of the Study:

• To review studies integrating multi-omics datasets for improved drug response prediction (DRP).
• To outline common omics types and computational approaches for DRP.
• To summarize challenges and future directions in multi-omics DRP.

Main Methods:

• Review of studies integrating multi-omics data for DRP.
• Categorization of omics types (e.g., genomics, transcriptomics).
• Analysis of computational approaches: classical machine learning (ML), deep learning, and multimodal integration frameworks.

Main Results:

• Identified key omics types and computational methods used in DRP.
• Detailed methodologies and evaluation metrics (e.g., AUC, F1 score, MSE).
• Summarized challenges in integrating high-dimensional, multimodal data.

Conclusions:

• Multi-omics data integration is crucial for advancing DRP in precision medicine.
• Advanced computational models are needed to handle complex omics data.
• Further research is required to overcome integration challenges and improve predictive accuracy.