The relationship between protein structure and function: a comprehensive survey with application to the yeast genome

Area of Science:

• Structural bioinformatics
• Proteomics
• Genomics

Background:

• Protein function prediction often relies on sequence comparison, but its reliability is not fully understood.
• Understanding the link between protein structure and function is crucial for accurate genomic annotation.

Purpose of the Study:

• To systematically investigate the relationship between protein structure and function.
• To assess the reliability of predicting protein function based on structural similarity.

Main Methods:

• Utilized the SCOP database for structural classification and the Enzyme Commission (EC) for functional classification of enzymes.
• Analyzed protein data across multiple genomic classifications including yeast, COGs, CATH, and MIPS.
• Developed a graph to quantify the probability of reliable functional annotation transfer based on sequence and structural similarity.

Main Results:

• Major SCOP fold classes exhibit distinct functional propensities; for example, alpha/beta folds are enriched in enzymes like transferases and hydrolases.
• Specific protein structural folds, such as TIM-barrel and Rossmann, are highly versatile, associating with numerous functions.
• Ancient proteins show a more even distribution of functions across structural classes compared to modern proteins.

Conclusions:

• Protein structure is a strong indicator of protein function, offering a more reliable basis for prediction than sequence similarity alone.
• Identifying versatile protein folds and their associated functions can improve genome annotation accuracy.
• The developed graph provides a quantitative tool for assessing the reliability of functional annotation transfer in bioinformatics.