New Score Assesses Protein-Protein Interfaces
The Topf group (HPI, UKE), at the Centre for Structural Systems Biology CSSB, has created a score to assess protein-protein interactions in the structures of macromolecular assemblies derived from cryo-EM maps.
To understand the complex interactions that occur within biological systems, researchers use various imaging methods to examine molecular assemblies and generate detailed structural models of the assemblies’ proteins. One such method is electron cryo-microscopy (cryoEM) which enables the imaging of large molecular complexes in their native environment. The images generated by cryoEM are, however, often not at the so-called high-resolution range and the resulting molecular models often contain several errors.
To help reduce these errors, the cryoEM community has been developing a “toolbox” of validation scores to evaluate the overall reliability of these molecular models. This “toolbox” has just been expanded with the addition of the Protein Interface-score (PI-score) developed by the Topf group. Recently published in Nature Communications, the PI-score is a validation score applied to specifically assess protein-protein interfaces, the location where two individual proteins physically interact in cryoEM structures.
Sony Malhotra, a former postdoc in the Topf group, systematically examined the protein-protein interfaces modeled in structures solved using high-resolution X-ray crystallography. “We have trained a machine learning based classifier to recognize native/native-like interfaces,” explains Malhotra “The added benefit of this score is that it not dependent on cryoEM map data.”
PI-scores are provided for all the publicly available cryoEM structures thus evaluating more than 5,000 interfaces including recent SARS-CoV-2 complexes. The PI-score was able to flag interface errors in several low and medium-resolution structures and in some instances the researchers were able to suggest refinements to the model based on the PI-score. “We hope that the PI-Score will become a powerful complementary assessment tool for the cryo-EM community and ultimately result in more accurate cryo-EM based models,” explains Topf.
The software to calculate the PI-score is freely available for academic use through: https://gitlab.com/topf-lab/pi_score.
Malhotra S, Joseph AP, Thiyagalingam J, Topf M (2021) Assessment of protein–protein interfaces in cryo-EM derived assemblies. Nat Commun 12:3399. https://doi.org/10.1038/s41467-021-23692-x