Feasibility in the inverse protein folding protocol.

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Science

Background:

• Protein structure (3D)-sequence (1D) compatibility evaluation methods, known as threading, have advanced significantly.
• Forward-folding search (fold recognition) predicts protein structure from sequence.
• Inverse-folding search, identifying homologous sequences from a given structure, presents a greater challenge.

Purpose of the Study:

• To investigate the feasibility and effectiveness of the inverse-folding search approach.
• To compare the performance of inverse-folding and forward-folding searches.
• To optimize the inverse-folding protocol for improved accuracy in homology detection.

Main Methods:

• Developed a structural library of approximately 400 well-resolved, dissimilar protein structures.
• Included 163 structures with known remote homologs within the library.
• Evaluated both forward- and inverse-folding search protocols using this library.

Main Results:

• The inverse-folding protocol demonstrated superior effectiveness compared to the forward-folding protocol after adjusting compatibility function reference states.
• Adjusting reference states had minimal impact on the forward-folding search's performance.
• Scoring based on direct 3D-1D alignment proved more effective for inverse-folding than re-mounted sequence scoring.

Conclusions:

• Optimized inverse-folding search is a more potent method for identifying homologous protein sequences from structures than forward-folding.
• The direct 3D-1D alignment score is crucial for effective inverse-folding.
• Findings have significant implications for protein structure prediction and homology analysis.