VGIchan

VGIchan predicts and classifies voltage-gated ion channels (VGICs) from primary amino acid sequences for identification and characterization of channel types.

Key Features:

• Prediction using amino acid composition: Support Vector Machine (SVM) models using amino acid composition achieve 82.89% accuracy for detecting VGICs.
• Prediction using dipeptide composition: SVM models using dipeptide composition achieve 85.56% accuracy for detecting VGICs.
• PSI-BLAST integration: Incorporation of a PSI-BLAST similarity search increases prediction accuracy from 85.56% to 89.11% by leveraging sequence similarity.
• Classification into channel types: An SVM method using dipeptide composition classifies VGICs into sodium, potassium, calcium, and chloride channels with 96.89% overall accuracy.
• Hybrid SVM–HMM classification: A hybrid approach combining dipeptide-based SVM with Hidden Markov Models (HMM) improves classification accuracy to 97.78%.

Scientific Applications:

• Novel channel identification: Sequence-based prediction facilitates discovery of previously unannotated voltage-gated ion channels from protein sequences.
• Channel-type annotation: Classification into sodium, potassium, calcium, and chloride channels supports functional annotation of ion channel sequences.
• Physiological and functional studies: Accurate sequence-based identification and typing of VGICs aids investigations of roles in nerve impulse transmission and muscle contraction.

Methodology:

Support Vector Machines (SVM) trained on amino acid composition and dipeptide composition; PSI-BLAST similarity search for enhanced prediction; Hidden Markov Models (HMM) combined with dipeptide-based SVM in a hybrid classification approach.