SUPRB

SUPRB predicts protein tertiary structures by threading target sequences onto template structures using suboptimal alignments and probabilistic contact potentials to improve template-based modeling.

Key Features:

• Suboptimal Alignments: Uses suboptimal alignments rather than only the top-scoring alignment, exploiting low-rank alignments to extract more accurate residue contact information.
• Probabilistic Contact Handling: Implements a probabilistic approach to integrate contact potentials derived from suboptimal alignments, outperforming partly thawed approaches that depend only on optimal alignments and simple re-ranking strategies.
• Integration with Modeller: Provides suboptimal alignments as input to Modeller to generate refined template-based structural models.

Scientific Applications:

• Computational Protein Structure Prediction: Improves accuracy of template-based protein structure prediction through enhanced contact inference from suboptimal alignments.
• Template Recognition for Remote Homologs: Facilitates identification and utilization of templates from structurally similar but remotely related proteins not easily detected by conventional homology searches.

Methodology:

Examines effects of suboptimal alignments in threading-based template selection, employs a probabilistic model of contact potentials derived from those alignments, and supplies alignments to Modeller for model building.