GSTPred

GSTPred predicts Glutathione S-transferase (GST) proteins from protein sequences to identify proteins involved in detoxification of exogenous and endogenous chemicals and in stress-response mechanisms.

Key Features:

• Dataset Utilization: Trained on a balanced dataset comprising 107 GST proteins and 107 non-GST proteins.
• Algorithm: Uses Support Vector Machines (SVM) for classification.
• Input Features: Analyzes amino acid composition, dipeptide composition, and tripeptide composition as sequence-derived features.
• Performance: Achieved prediction accuracies of 91.59% (amino acid composition), 95.79% (dipeptide composition), and 97.66% (tripeptide composition), with HMM-based searching reported at 96.26%.
• Validation Technique: Evaluated using five-fold cross-validation.

Scientific Applications:

• Detoxification and stress-response studies: Identification of GST proteins to investigate cellular mechanisms for detoxification and survivability under stress.
• Target identification and pathway analysis: Support for locating potential therapeutic targets and for analyzing biochemical pathways involved in cellular defense mechanisms.

Methodology:

Support Vector Machines (SVM) trained on a balanced dataset of 107 GST and 107 non-GST proteins using amino acid, dipeptide, and tripeptide composition features, evaluated by five-fold cross-validation, with reported accuracies and comparison to HMM-based searching.