Accelerating the Discovery of Anticancer Peptides through Deep Forest Architecture with Deep Graphical Representation

Area of Science:

• Biotechnology
• Bioinformatics
• Computational Biology

Background:

• Cancer remains a leading global health threat, driving the need for innovative treatments.
• Peptide-based therapies are gaining prominence in cancer research.
• Accurate prediction of anticancer peptides (ACPs) is vital for developing new cancer therapies.

Purpose of the Study:

• To propose and evaluate a novel machine learning framework, GRDF, for the precise identification of ACPs.
• To leverage deep graphical representation and deep forest architecture for enhanced ACP prediction.
• To provide an interpretable and robust model for ACP discovery.

Main Methods:

• Developed the GRDF framework integrating deep graphical representation and deep forest architecture.
• Extracted graphical features from peptide physicochemical properties and evolutionary information.
• Employed a layer-by-layer cascade deep forest algorithm for model construction.

Main Results:

• GRDF achieved state-of-the-art performance on two benchmark datasets (Set 1 and Set 2).
• Achieved 77.12% accuracy and 77.54% F1-score on Set 1, and 94.10% accuracy and 94.15% F1-score on Set 2.
• Demonstrated superior robustness and interpretability compared to baseline methods.

Conclusions:

• The GRDF framework is highly effective for identifying ACPs.
• GRDF can significantly facilitate the discovery of novel anticancer peptides.
• This approach holds promise for advancing the development of new cancer treatments.