iNP_ESM: Neuropeptide Identification Based on Evolutionary Scale Modeling and Unified Representation Embedding...

Area of Science:

• Biochemistry and Molecular Biology
• Neuroscience
• Computational Biology

Background:

• Neuropeptides are vital biomolecules for physiological functions, particularly in the nervous system.
• Accurate neuropeptide identification is crucial for understanding neural regulation.
• Traditional methods for neuropeptide analysis are resource-intensive, driving the need for advanced computational tools.

Purpose of the Study:

• To develop a highly accurate and efficient machine learning model for neuropeptide identification.
• To integrate cutting-edge protein language models for enhanced prediction capabilities.
• To establish a benchmark for future neuropeptide prediction models.

Main Methods:

• Construction of a Support Vector Machine (SVM)-based predictor, iNP_ESM.
• Integration of Evolutionary Scale Modeling (ESM) and Unified Representation (UniRep) protein language models.
• Application of feature fusion and feature selection strategies for model optimization.
• Validation using Uniform Manifold Approximation and Projection (UMAP) visualization.

Main Results:

• iNP_ESM achieved high prediction accuracy, with 0.937 in cross-validation and 0.928 in independent testing.
• The model demonstrated superior performance compared to existing neuropeptide prediction methods.
• Optimization strategies significantly improved prediction accuracy.

Conclusions:

• The iNP_ESM model offers a powerful and accurate tool for neuropeptide recognition.
• This advancement has significant implications for neurological disease research and clinical applications.
• The model is poised for broader use with future improvements in neuropeptide data and computational techniques.