Database algorithm for generating protein backbone and side-chain co-ordinates from a C alpha trace application to...

Area of Science:

• Computational Biology
• Structural Biology
• Biophysics

Background:

• Protein structure determination is complex, involving techniques like crystallography and NMR spectroscopy.
• Accurate protein models are crucial for structure determination, homology modeling, and protein design.
• Generating all-atom protein models from simplified representations (e.g., C-alpha trace) remains a challenge.

Purpose of the Study:

• To develop an automated procedure for generating full protein atomic coordinates from a C-alpha trace and amino acid sequence.
• To provide a computational tool that assists in protein structure determination and design.

Main Methods:

• A protein structure database is searched for fragments matching the C-alpha trace.
• A best path algorithm selects optimal backbone segments with minimal joint mismatches.
• Side-chains are added using a Monte Carlo procedure with simulated annealing and a rotamer library.

Main Results:

• Backbone reconstruction achieved 0.4-0.6 Å root-mean-square position deviation, with minimal peptide flips.
• Side-chain atom positions in the protein core showed 1.6 Å accuracy (r.m.s. deviation).
• 70% of chi 1 angles were within 30 degrees of experimental structures for X-ray structures >2.5 Å resolution.

Conclusions:

• The developed method accurately generates protein backbone and side-chain coordinates.
• The accuracy is sufficient for identifying potential errors in existing protein coordinate sets.
• The MaxSprout program offers a valuable tool for computational protein structure analysis and design.