ACP-CLB: An Anticancer Peptide Prediction Model Based on Multichannel Discriminative Processing and Integration of...

Area of Science:

• Bioinformatics
• Computational Biology
• Drug Discovery

Background:

• Anticancer peptides (ACPs) show promise as alternatives to conventional cancer treatments.
• Wet-lab validation of ACPs is resource-intensive.
• Current AI methods for ACP identification often use a one-size-fits-all approach to feature processing.

Purpose of the Study:

• To develop an advanced AI framework for more accurate and efficient identification and classification of anticancer peptides.
• To address the limitations of uniform feature processing in existing ACP identification models.

Main Methods:

• Proposed a multichannel discriminative processing framework using distinct neural networks for various feature types.
• Integrated Large Pretrained Protein Language Models to extract deep sequence features.
• Evaluated the framework against state-of-the-art models on four diverse datasets.

Main Results:

• Achieved significant performance improvements across most evaluation metrics compared to existing methods.
• Demonstrated superior accuracy in distinguishing and identifying anticancer peptides.
• Validated the effectiveness of specialized processing for different feature types.

Conclusions:

• The proposed framework offers a more effective approach to anticancer peptide identification.
• Highlights the importance of tailored feature processing for enhancing AI model performance in bioinformatics.
• Provides a valuable tool for researchers in the field of anticancer peptide discovery.