MDTL-ACP: Anticancer Peptides Prediction Based on Multi-Domain Transfer Learning.

Area of Science:

• Biochemistry
• Computational Biology
• Drug Discovery

Background:

• Anticancer peptides (ACPs) show promise for cancer therapy due to broad activity and low resistance.
• Traditional ACP discovery is costly and time-consuming.
• Existing computational methods are limited by scarce ACP data.

Purpose of the Study:

• To develop an effective computational model for identifying novel anticancer peptides.
• To address the limitations of scarce data in ACP discovery.

Main Methods:

• Proposed MDTL-ACP, a multi-domain transfer learning model.
• Utilized abundant antimicrobial peptides (AMPs) from four domains as source data.
• Employed artificial neural networks for feature extraction and classification.

Main Results:

• MDTL-ACP successfully discriminates novel ACPs from inactive peptides.
• Transfer learning enhanced prediction performance in the target ACP domain.
• The model outperformed existing ACP prediction methods.

Conclusions:

• Multi-domain transfer learning is a viable strategy for enhancing ACP prediction.
• MDTL-ACP offers a more efficient approach to discovering potential anticancer peptides.
• This method can overcome data scarcity challenges in computational drug discovery.