SDP

SDP predicts real inter-residue distances to improve protein structure prediction by accurately modeling large and sequence-distant residue separations.

Key Features:

• Feature Optimization: Reduces the feature set to two coevolutionary and three non-coevolutionary types to minimize noise and enhance prediction accuracy.
• Deep learning framework: Uses a deep learning model that integrates the selected features to predict real inter-residue distances.
• Improved Prediction Accuracy: Achieves at least a 10% increase in mean Local Distance Difference Test (LDDT) scores on benchmark protein datasets compared to state-of-the-art methods.
• Efficient Conformational Search: Produces more accurate inter-residue distance predictions that facilitate more efficient conformational searches during protein structure prediction.

Scientific Applications:

• Structural biology: Supports development of accurate 3D protein models for understanding protein function, interactions, and mechanisms.
• Bioinformatics research: Serves in benchmarking and improving protein structure prediction methods using inter-residue distance evaluation.
• Drug discovery: Provides distance constraints useful for structure-based drug design and ligand modeling.
• Enzyme design: Aids enzyme engineering by supplying precise inter-residue distance information for model building.

Methodology:

Employs a deep learning framework that integrates a carefully selected set of features—two coevolutionary and three non-coevolutionary—to predict real inter-residue distances and reduce noise through feature selection.