ACP-DRL: an anticancer peptides recognition method based on deep representation learning.

Area of Science:

• Biochemistry
• Computational Biology
• Oncology

Background:

• Cancer is a leading cause of mortality globally, necessitating novel therapeutic strategies.
• Anticancer peptides (ACPs) show promise in inhibiting tumor growth with fewer side effects than traditional treatments.
• Current methods for identifying ACPs via wet-lab experiments are inefficient and costly.

Purpose of the Study:

• To introduce ACP-DRL, a deep representation learning-based method for accurate and efficient recognition of anticancer peptides.
• To overcome the limitations of traditional wet-lab identification methods for ACPs.
• To leverage advanced computational techniques for accelerating the discovery of novel ACPs.

Main Methods:

• Integration of protein language models with in-domain further pre-training for enhanced representation learning.
• Utilization of bidirectional long short-term memory (BiLSTM) networks to extract sequence-based amino acid features.
• Development of a deep learning framework (ACP-DRL) for ACP recognition, independent of sequence length and manual feature engineering.

Main Results:

• ACP-DRL demonstrates superior performance compared to existing ACP recognition methods.
• The model effectively extracts relevant features from amino acid sequences without manual intervention.
• Achieved high accuracy in identifying potential anticancer peptides, reducing experimental costs and time.

Conclusions:

• ACP-DRL offers a computationally efficient and cost-effective alternative for identifying anticancer peptides.
• The integration of protein language models and deep learning represents a significant advancement in computational drug discovery for cancer.
• This method facilitates the accelerated discovery and development of novel peptide-based cancer therapeutics.