SPOT-Disorder2

SPOT-Disorder2 predicts intrinsic disorder in protein sequences and disordered regions to identify functional disordered regions relevant to protein function and disease.

Key Features:

• Ensemble Deep Learning Architecture: Integrates SE-ResNeXt (Squeeze-and-Excitation residual inception) and long short-term memory (LSTM) networks and leverages evolutionary information and predicted one-dimensional structural properties for disorder prediction.
• Improved Accuracy: Demonstrates substantial improvements over an LSTM-only predecessor and outperforms other state-of-the-art methods on independent datasets with differing ratios of disordered to ordered amino acid residues and varying levels of evolutionary information.
• Semi-Disordered Region Prediction: Accurately predicts semi-disordered regions and aids identification of molecular recognition features (MoRFs), surpassing some MoRF-specific methods.

Scientific Applications:

• Protein function and structure annotation: Identification of functional disordered regions for interpretation of protein function and structural dynamics.
• Molecular recognition and signaling: Detection of MoRFs and other interaction sites involved in molecular interactions and signaling pathways.
• Disease mechanism analysis: Characterization of intrinsically disordered proteins (IDPs) and intrinsically disordered regions (IDRs) implicated in disease mechanisms and biomarkers.
• Drug discovery and target identification: Prioritization of disordered regions and interaction sites relevant to therapeutic target development.

Methodology:

Ensembles of SE-ResNeXt (Squeeze-and-Excitation residual inception) and LSTM networks are trained using evolutionary information and predicted one-dimensional structural properties and evaluated on independent datasets with varying disorder-to-order ratios and evolutionary information levels.