neuropipred

neuropipred predicts and designs insect neuropeptides to identify candidates with enhanced efficacy for pest control in agricultural settings.

Key Features:

• Amino acid composition and positional preferences: Comprehensive analysis identified overall residue preferences for C, D, E, F, G, N, S, and Y, N-terminal preferences for A, N, F, D, P, S, and I, and C-terminal preferences for L, R, P, F, N, and G.
• Input features: Models use amino acid composition, dipeptide composition, and binary profiles as sequence-derived features.
• Machine learning models: Various machine learning techniques were evaluated, with a Support Vector Machine (SVM) built on dipeptide composition emphasized as the best-performing approach.
• Performance metrics: NeuroPIpred_DS1 achieved 86.50% accuracy and 0.73 MCC on training and 83.71% accuracy and 0.67 MCC on validation; NeuroPIpred_DS2 achieved 97.47% accuracy and 0.95 MCC on training and 97.93% accuracy and 0.96 MCC on validation.

Scientific Applications:

• Pest-control peptide prediction: Identification of insect neuropeptide sequences likely to affect pest physiology for agricultural pest management.
• Peptide design and optimization: Design and selection of modified neuropeptides with enhanced efficacy against target pest populations.
• Sequence determinant analysis: Characterization of residue and positional preferences to inform mechanistic understanding and design rules.

Methodology:

Analysis of amino acid composition and positional residue preferences, extraction of features (amino acid composition, dipeptide composition, binary profiles), and training of machine learning models including Support Vector Machine on dipeptide composition.