SSpro

SSpro predicts protein secondary structure and relative solvent accessibility by combining neural network–based ab initio prediction with homology analysis to improve accuracy for protein 3D structure prediction and functional inference.

Key Features:

• Hybrid Approach: Combines neural network ab initio prediction with homology-based analysis to leverage both sequence-derived profiles and structural similarities.
• Modular Predictors: Provides SSpro and ACCpro predictors and their multi-class variants SSpro8 and ACCpro20 for secondary structure and solvent accessibility prediction.
• Sequence Similarity Utilization: Uses sequence profiles for sequence-similarity-based predictions and distinguishes this from sequence-based structural similarity that leverages matches to Protein Data Bank entries.
• High Accuracy Rates: Achieves ~79–80% accuracy for SSpro and ~79% for ACCpro using sequence similarity alone, and increases to 92.9% (SSpro) and 90% (ACCpro) when incorporating sequence-based structural similarity.
• Technical Challenges Addressed: Targets open challenges such as attaining ≥80% accuracy without known protein similarities and reaching ≥85% accuracy using sequence similarity alone.

Scientific Applications:

• Protein structure prediction: Supports secondary structure and solvent accessibility annotation used in protein 3D structure modeling.
• Structural biology: Facilitates interpretation of protein folding, dynamics, and residue exposure in experimental and computational studies.
• Protein evolution and function: Aids comparative analyses of sequence-structure relationships for evolutionary and functional inference.
• Drug design and functional annotation: Provides structural context for identifying functional sites and guiding structure-based drug discovery.

Methodology:

Employs neural network–based ab initio prediction and homology analysis incorporating sequence profiles and sequence-based structural similarity to Protein Data Bank entries.