SA-MTP: a structure-aware framework for multifunctional therapeutic peptide annotation.

Area of Science:

• Biochemistry
• Computational Biology
• Drug Discovery

Background:

• Therapeutic peptides are promising drug candidates due to diverse biological activities.
• Accurate functional annotation of therapeutic peptides is challenging due to short sequences, structural flexibility, and multifunctionality.

Purpose of the Study:

• To introduce a novel framework, Structure-Aware Multi-Label Therapeutic Peptide Predictor (SA-MTP), for multifunctional annotation of therapeutic peptides.
• To improve the accuracy of predicting biological functions for therapeutic peptides.

Main Methods:

• SA-MTP integrates pretrained protein language models with a graph attention network.
• It captures sequence semantics and probabilistic structural features using input-dependent structure-aware graphs.
• The framework accounts for conformational variations common in short peptides.

Main Results:

• SA-MTP demonstrated superior performance compared to existing methods across 15 therapeutic function categories.
• The model achieved higher accuracy, F1-score, and Matthews correlation coefficient.
• The structure-aware approach effectively addresses challenges in peptide annotation.

Conclusions:

• SA-MTP provides a robust and accurate method for the multifunctional annotation of therapeutic peptides.
• This advancement can accelerate the drug development process by improving the understanding of peptide functions.
• The framework offers a significant improvement in predicting therapeutic peptide activities.