NeuroPID

NeuroPID identifies neuropeptide precursors (NPPs) and predicts secreted neuromodulators across metazoan species to enable discovery of neuropeptides from protein or transcript sequence sets.

Key Features:

• Machine Learning-Based Identification: NeuroPID employs support vector machines and ensemble decision trees trained on 1,418 NPPs annotated in UniProtKB, reporting 89–94% accuracy and 90–93% precision.
• Sequence Feature Extraction: The tool extracts numerous sequence-based features that capture biophysical and informational-statistical properties unique to NPPs.
• Multi-classifier Ranked Predictions: For large input sets NeuroPID provides a ranked list of predictions from four machine-learning classifiers, highlighting potential neuropeptide precursors and secreted cell modulators enriched in cleavage sites.
• Discovery and Prediction Utility: NeuroPID is applicable to unannotated transcriptomes and mass spectrometry data for identifying and prioritizing novel NPP candidates relevant to behavioral, physiological, and cellular modulation studies.

Scientific Applications:

• Identification of Uncharacterized NPPs: NeuroPID facilitates discovery of previously uncharacterized neuropeptide precursors across metazoan species by predicting candidate sequences.
• Target Prioritization for Functional Studies: NeuroPID provides prioritized candidate sequences for downstream investigation of neuropeptide roles in metabolism, sensation, behavior, and other biological processes.

Methodology:

NeuroPID trains machine-learning models (support vector machines and ensemble decision trees) on annotated NPP data from UniProtKB, extracts sequence-based biophysical and informational-statistical features, and outputs ranked predictions from four classifiers for large sequence sets.