Searching the protein structure database for ligand-binding site similarities using CPASS v.2

Area of Science:

• Biochemistry
• Structural Biology
• Bioinformatics

Background:

• Approximately 30% of ~8.5 million protein sequences in NCBI RefSeq are hypothetical or uncharacterized.
• The Comparison of Protein Active-Site Structures (CPASS) database and software infer protein function from ligand binding site characteristics.
• CPASS is integral to the Functional Annotation Screening Technology by NMR (FAST-NMR) protocol for annotating unknown proteins.

Purpose of the Study:

• To report a major upgrade to the CPASS software and database (CPASS v.2).
• To enhance the utility, flexibility, and speed of CPASS for functional annotation.
• To improve the accuracy of functional inference for proteins with unknown functions.

Main Methods:

• CPASS v.2 features a layered architecture for flexibility and portability, enabling distributed computing over the Open Science Grid (OSG).
• Enhanced user interface allows automatic or manual definition of ligand-binding sites and flexible comparison modes (pair-wise, one vs. all, etc.).
• Incorporation of solvent accessible surface area, ligand root-mean square difference, and Cβ distances into the CPASS similarity function.

Main Results:

• CPASS v.2 is over 10x faster than the previous version and supports multiple simultaneous job submissions.
• The CPASS database size increased by ~38%, improving the probability of successful searches.
• Modified similarity function reduced false positive scores by ~30% without affecting true positives.

Conclusions:

• Receiver operating characteristic (ROC) curves confirm a strong correlation between CPASS similarity scores and accurate functional assignments.
• CPASS v.2 significantly enhances the speed and accuracy of inferring protein function from structural data.
• A CPASS similarity score ≥ 30% indicates functional similarity, aiding in the annotation of hypothetical proteins.