Evaluation of comparative protein structure modeling by MODELLER-3

Area of Science:

• Structural bioinformatics
• Computational biology
• Protein structure prediction

Background:

• Homology modeling is a key technique for predicting protein structures.
• Comparative modeling relies on accurate sequence alignments and template selection.
• Assessing model accuracy is crucial for understanding protein function.

Purpose of the Study:

• To evaluate the accuracy of homology-derived 3D protein models for DFR1, PTE2A3, and UBC9(24).
• To compare current model quality with previous assessments from the Critical Assessment of Techniques for Protein Structure Prediction (CASP) meetings.
• To identify persistent sources of error in homology modeling.

Main Methods:

• Comparative modeling using homology-derived 3D structures.
• Evaluation of models against experimental X-ray structures.
• Analysis of sequence-structure alignments and identification of errors.

Main Results:

• DFR1 and PTE2A3 models showed slight improvements over template structures.
• UBC9(24) model accuracy was hampered by N-terminal helix misalignment.
• Major error types in side chain packing and segment conformation persisted, similar to 1994.
• Iterative refinement improved alignment accuracy, particularly for DFR C-terminus.

Conclusions:

• Homology modeling accuracy has modestly improved due to refined methods.
• Iterative model building and alignment correction are effective strategies.
• Persistent challenges in homology modeling necessitate continued methodological development.