PredGPI

PredGPI predicts Glycosylphosphatidylinositol (GPI)-anchored proteins and the C-terminal omega-sites in eukaryotic protein sequences.

Key Features:

• Dual-model approach: Integrates a Hidden Markov Model (HMM) with a Support Vector Machine (SVM) to detect GPI anchors and locate omega-sites in protein sequences.
• Omega-site and C-terminal annotation: Identifies the proteolytic omega-site and the subsequent GPI attachment at the C-terminal residue.
• Curated training dataset: Trained on a non-redundant dataset of experimentally validated GPI-anchored proteins.
• Performance evaluation: Shows a lower false positive rate compared to earlier methods and has been validated against high-throughput experimental results.

Scientific Applications:

• Proteome annotation: Annotation of whole eukaryotic proteomes for GPI-anchored proteins and omega-site positions.
• Protein localization and function studies: Support studies of protein localization, post-translational modification, and membrane association.
• Cell signaling and host–pathogen research: Aid investigations into cell signaling, pathogen–host interactions, and membrane biology where GPI-anchored proteins are involved.

Methodology:

Integrates an HMM with an SVM and is trained on a curated non-redundant dataset of experimentally validated GPI-anchored proteins.