DiMoVo: a Voronoi tessellation-based method for discriminating crystallographic and biological protein–protein interactions

Abstract

Software is available at http://fifi.ibbmc.u-psud.fr/DiMoVo.

Structure of the Mature Streptococcal Cysteine Protease Exotoxin mSpeB in Its Active Dimeric Form

“…Independently, the DiMoVo score is 0.67, suggesting a biologically relevant complex (score, N0.5). 18,22 The other interface scores in the crystal give DiMoVo values around zero.…”

An accurate classification of native and non-native protein-protein interactions using supervised and semi-supervised learning approaches

“…As the combined performance of our model surpasses many of the previously developed classifiers [7], it shows that the defined features are capable of capturing the difference between the physiological interfaces and crystal-packing ones.…”

“…Using the interface size and residue conservation has allowed to reach the accuracy of 98.3% in predicting the crystal-packing interfaces for a set of homodimers [6]. Recent methods have been able to successfully distinguish between the crystal packing and biological interface, with the accuracies ranging from 76% to 97% and recalls ranging from 68% to 97% (methods with higher accuracies tend to have lower recall values) [7,8].…”

Comparing Voronoi and Laguerre Tessellations in the Protein-Protein Docking Context

“…In this construction, all the sites, in our case all the amino acids, are considered equivalent. This modeling lead us to an accurate scoring function for protein-protein docking [14,15], and allowed us to efficiently distinguish biological and crystallographic dimers [20]. However, the mean volumes of Voronoi cells computed using this construction are very different from the volumes that can be computed from an atomic-Voronoi construction, and considered as reference volumes (Figure 1 top).…”