PSIONplus

PSIONplus predicts ion channels, their types, and four major subtypes of voltage-gated channels from protein sequence data to support annotation and characterization of membrane proteins.

Key Features:

• Comprehensive Prediction Scope: Predicts the presence of ion channels and classifies them into channel types and the four major subtypes of voltage-gated channels.
• Sophisticated Machine Learning Model: Employs a support vector machine (SVM) that integrates BLAST sequence similarity searches with inputs including evolutionary profiles, predicted secondary structure, solvent accessibility, and intrinsic disorder.
• Innovative Input Types: Uses evolutionary profiles as a predictive input, which empirical evaluation identified as the most significant contributor among considered input types.
• High Predictive Performance: Reports accuracies of 85.4% for ion-channel detection, 68.3% for channel-type classification, and an average accuracy of 96.4% for discriminating among the four voltage-gated channel subtypes.

Scientific Applications:

• Drug Discovery: Identifies ion channels as potential drug targets to inform pharmacological research.
• Genomic Research: Annotates genomes with ion channel information to support studies of genetic contributions to physiological functions.
• Structural Biology: Supports prediction of structural features and functional properties of ion channels to guide experimental design and hypothesis generation.

Methodology:

Combines BLAST-based sequence similarity searches with a support vector machine that integrates evolutionary profiles, predicted secondary structure, solvent accessibility, and intrinsic disorder.