DictyOGlyc

DictyOGlyc predicts N-acetylglucosaminyl (GlcNAc) O-glycosylation acceptor sites in Dictyostelium discoideum proteins, identifying potential alpha-linked GlcNAc attachments to hydroxyl groups in secreted and membrane proteins to support glycobiology research.

Key Features:

• Artificial Neural Networks: A jury of artificial neural networks trained on 39 experimentally validated O-linked GlcNAc sites from D. discoideum glycoproteins recognizes sequence context and protein surface accessibility.
• High Accuracy: Cross-validation results reported that 97% of glycosylated and non-glycosylated sites were correctly identified.
• Sequence Context Analysis: The data show abundant periodicity of proline alternating with hydroxyl amino acids at positions -3, -1, +1, +3, etc., suggesting glycosylated residues are often situated within beta-strands at even positions relative to each other.
• Proteome Scanning: The method can scan and rank a significant portion of the Dictyostelium proteome from public databases, predicting that 25–30% of proteins may be glycosylated and identifying acceptor sites including previously unmapped locations.
• Functional and Cellular Compartment Classification: Predicted glycosylation sites are classified by functional category and cellular compartment to reveal preferential glycosylation patterns across contexts.

Scientific Applications:

• Novel Site Identification: Guides identification of novel O-GlcNAc acceptor sites in D. discoideum proteins, including previously unmapped locations.
• Functional Annotation: Supports interpretation of protein function and interactions by indicating sites of potential glycosylation that may affect protein behavior.
• Proteome-wide Pattern Analysis: Enables analysis of glycosylation distribution across the Dictyostelium proteome, including estimates that 25–30% of proteins may be glycosylated.
• Experimental Design and Hypothesis Generation: Provides high-accuracy predictions to inform experimental design and hypothesis generation in studies of cell signaling, immune response, and disease mechanisms.

Methodology:

A jury of artificial neural networks was trained on 39 experimentally validated O-linked GlcNAc sites from D. discoideum glycoproteins using sequence context and protein surface accessibility as inputs, incorporating observed proline/hydroxyl residue periodicity and structural considerations, validated by cross-validation, and applied to scan and rank the Dictyostelium proteome.