Structural difficulty index: a reliable measure for modelability of protein tertiary structures

Area of Science:

• Structural Biology
• Bioinformatics
• Computational Biology

Background:

• Protein tertiary-structure prediction success is often based on sequence similarity to known structures.
• This traditional measure may not accurately reflect a protein's true modelability.
• A more robust method is needed to assess prediction feasibility.

Purpose of the Study:

• To introduce a novel 'structural difficulty' (SD) index for evaluating protein structure prediction.
• To assess the potential for developing large-scale, high-quality proteome structural databases.
• To improve the accuracy of protein structure modeling and its applications.

Main Methods:

• Developed the structural difficulty (SD) index using secondary structures, homology, and physicochemical features.
• Applied the SD index to assess the modelability of human and viral proteomes.
• Calculated the percentage of proteins with reliable quality structures within a 3 Å root mean square deviation.

Main Results:

• The SD index provides a more reliable measure of protein modelability than sequence similarity alone.
• Approximately 37% of manually curated human soluble proteins and 64% of UniProtKB soluble proteins are modelable.
• For human pathogenic viruses, 1336 out of 2365 globular viral proteins were identified as modelable.

Conclusions:

• The SD index is a valuable tool for assessing protein structure prediction accuracy.
• It facilitates the development of species-wise structural proteomic databases.
• Reliable protein structures aid in accelerating function annotation and drug development.