NeuroCS

NeuroCS predicts cleavage sites in neuropeptide precursors to support identification of neuropeptides and the study of neuronal regulation.

Key Features:

• Cleavage site prediction: Predicts cleavage sites in neuropeptide precursors using computational models.
• Data sources: Uses neuropeptide precursor sequences sourced from SwissProt and NueoPedia and partitions them into training and independent testing datasets.
• Feature extraction: Employs six distinct feature extraction schemes to generate multiple feature sets.
• Feature selection: Applies feature selection methods to identify optimal feature subsets for each extraction scheme.
• Modeling algorithm: Constructs predictive models using support vector machines (SVM) trained on selected feature subsets.
• Model variants: Develops six predictive models corresponding to different feature types, including an enhanced amino acid composition-based model.
• Performance evaluation: Evaluates models using 5-fold cross-validation and an independent testing dataset.
• Performance metrics: The enhanced amino acid composition-based model achieved 91.60% cross-validation accuracy, 90.37% independent-test accuracy, and an Area under ROC curve of 0.9576.

Scientific Applications:

• Neuropeptide cleavage site annotation: Enables identification of cleavage sites within neuropeptide precursors for downstream biochemical and proteomic analyses.
• Neuroscience research: Facilitates studies of neuronal regulation and brain science by providing predicted cleavage sites to inform experimental and computational investigations.

Methodology:

Neuropeptide precursor sequences were sourced from SwissProt and NueoPedia and split into training and independent testing datasets; six feature extraction schemes were applied and optimal subsets selected via feature selection methods; support vector machines (SVM) were trained on the selected features and evaluated with 5-fold cross-validation and an independent test set.