DeepTPpred: A Deep Learning Approach With Matrix Factorization for Predicting Therapeutic Peptides by Integrating...

Area of Science:

• Biotechnology
• Computational Biology
• Drug Discovery

Background:

• Antibiotic resistance necessitates novel therapeutic strategies.
• Efficient prediction of therapeutic peptides is crucial for drug discovery.
• Existing methods lack comprehensive feature integration, particularly sequence length.

Purpose of the Study:

• To propose DeepTPpred, a novel deep learning approach for therapeutic peptide prediction.
• To integrate sequence length information as a distinct feature in therapeutic peptide prediction.
• To enhance the accuracy and scope of therapeutic peptide identification.

Main Methods:

• Developed a deep learning model incorporating matrix factorization for feature extraction.
• Embedded sequence length information alongside encoded amino acid sequences.
• Utilized a neural network with a self-attention mechanism for prediction.
• Validated the model on eight diverse therapeutic peptide datasets, including integrated and functional datasets.

Main Results:

• DeepTPpred demonstrated excellent prediction performance across multiple datasets.
• The integration of sequence length information proved effective.
• The model showed robust performance on updated ACP and CPP datasets.
• Experimental results confirm the efficacy of DeepTPpred in identifying therapeutic peptides.

Conclusions:

• DeepTPpred offers an effective and novel approach for therapeutic peptide prediction.
• Integrating sequence length information significantly improves prediction accuracy.
• This method holds promise for accelerating peptide drug discovery and addressing antibiotic resistance.