dprot

dprot predicts structurally disordered proteins and disordered regions to support molecular-level studies and characterization of their roles and associations with disease.

Key Features:

• Support Vector Machine models: Uses Support Vector Machine (SVM) models trained on sequence and profile compositions as input features.
• Amino acid composition module: An SVM module based on amino acid composition achieved sensitivity 75.6% and MCC 0.75.
• Dipeptide composition module: An SVM module based on dipeptide composition achieved sensitivity 73.2% and MCC 0.60.
• Secondary structure content: Incorporation of predicted secondary structure content (coil, sheet, helix) increased sensitivity to 76.8% and MCC to 0.77.
• Evolutionary information: Integration of evolutionary information from multiple sequence alignment profiles raised sensitivity to 78% and MCC to 0.78.
• Independent evaluation: Evaluation on an independent dataset of partially disordered proteins yielded a correct prediction rate of 86.6%.

Scientific Applications:

• Prediction of disordered regions: Identification of fully and partially disordered regions within proteins from sequence and profile data.
• Structure–function analysis: Annotation of disorder to study protein structure–function relationships.
• Disease association studies: Characterization of disorder in proteins implicated in disease mechanisms.

Methodology:

Support Vector Machine (SVM) models trained on sequence and profile compositions with modules using amino acid and dipeptide compositions, predicted secondary structure content (coil, sheet, helix), and evolutionary information from multiple sequence alignment profiles.