Discriminating physiological from non-physiological interfaces in structures of protein complexes: A community-wide study

Hugo Schweke, Qifang Xu, Gerardo Tauriello, Lorenzo Pantolini, Torsten Schwede, Frédéric Cazals, Alix Lhéritier, Juan Fernandez-Recio, Luis Angel Rodríguez-Lumbreras, Ora Schueler-Furman, Julia K. Varga, Brian Jiménez-García, Manon F. Réau, Alexandre M. J. J. Bonvin, Castrense Savojardo, Pier-Luigi Martelli, Rita Casadio, Jérôme Tubiana, Haim J. Wolfson, Romina OlivaDidier Barradas-Bautista, Tiziana Ricciardelli, Luigi Cavallo, Česlovas Venclovas, Kliment Olechnovič, Raphael Guerois, Jessica Andreani, Juliette Martin, Xiao Wang, Genki Terashi, Daipayan Sarkar, Charles Christoffer, Tunde Aderinwale, Jacob Verburgt, Daisuke Kihara, Anthony Marchand, Bruno E. Correia, Rui Duan, Liming Qiu, Xianjin Xu, Shuang Zhang, Xiaoqin Zou, Sucharita Dey, Roland L. Dunbrack, Emmanuel D. Levy, Shoshana J Wodak

Research output: Contribution to journalArticlepeer-review

4 Scopus citations

Abstract

Reliably scoring and ranking candidate models of protein complexes and assigning their oligomeric state from the structure of the crystal lattice represent outstanding challenges. A community-wide effort was launched to tackle these challenges. The latest resources on protein complexes and interfaces were exploited to derive a benchmark dataset consisting of 1677 homodimer protein crystal structures, including a balanced mix of physiological and non-physiological complexes. The non-physiological complexes in the benchmark were selected to bury a similar or larger interface area than their physiological counterparts, making it more difficult for scoring functions to differentiate between them. Next, 252 functions for scoring protein-protein interfaces previously developed by 13 groups were collected and evaluated for their ability to discriminate between physiological and non-physiological complexes. A simple consensus score generated using the best performing score of each of the 13 groups, and a cross-validated Random Forest (RF) classifier were created. Both approaches showed excellent performance, with an area under the Receiver Operating Characteristic (ROC) curve of 0.93 and 0.94, respectively, outperforming individual scores developed by different groups. Additionally, AlphaFold2 engines recalled the physiological dimers with significantly higher accuracy than the non-physiological set, lending support to the reliability of our benchmark dataset annotations. Optimizing the combined power of interface scoring functions and evaluating it on challenging benchmark datasets appears to be a promising strategy.
Original languageEnglish (US)
JournalPROTEOMICS
DOIs
StatePublished - Jun 27 2023

ASJC Scopus subject areas

  • Biochemistry
  • Molecular Biology

Fingerprint

Dive into the research topics of 'Discriminating physiological from non-physiological interfaces in structures of protein complexes: A community-wide study'. Together they form a unique fingerprint.

Cite this