A hybrid deep forest-based method for predicting synergistic drug combinations

Area of Science:

• Computational biology
• Bioinformatics
• Machine learning in drug discovery

Background:

• Drug combination therapy shows promise for complex diseases but faces challenges due to the vast search space for synergistic combinations.
• Experimental screening of potential drug combinations is time-consuming and resource-intensive.

Purpose of the Study:

• To develop an accurate and efficient computational method for predicting synergistic drug combinations in cancer cell lines.
• To address the challenges of imbalanced and high-dimensional data inherent in drug combination datasets.

Main Methods:

• Proposed ForSyn, an improved deep forest-based method incorporating two novel forest types.
• ForSyn is designed to handle medium-/small-scale, imbalanced, and high-dimensional datasets.
• Evaluated ForSyn against 12 state-of-the-art methods across eight diverse datasets.

Main Results:

• ForSyn achieved superior performance, ranking first on four key metrics compared to existing methods.
• Systematic analysis identified optimal configuration parameters for ForSyn.
• Experimental validation confirmed the predictive accuracy of ForSyn for novel drug combinations.

Conclusions:

• ForSyn offers a robust and effective approach for predicting synergistic drug combinations, accelerating drug discovery.
• Feature importance analysis within ForSyn can identify key genes potentially involved in cancer progression and therapeutic response.