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 Oliva; Didier 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

doi:10.1002/pmic.202200323

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^*

^*Corresponding author for this work

Research output: Contribution to journal › Article › Academic › peer-review

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 language	English
Article number	2200323
Number of pages	21
Journal	Proteomics
Volume	23
Issue number	17
Early online date	27 Jun 2023
DOIs	https://doi.org/10.1002/pmic.202200323
Publication status	Published - Sept 2023

Bibliographical note

Publisher Copyright:
© 2023 Wiley-VCH GmbH.

Funding

This study (SJW coordinator) was performed as part of the Implementation Plan (2019‐2021) of Activity‐2 of the 3DBioInfo Elixir Community ( https://elixir‐europe.org/communities/3d‐bioinfo ). We thank Alexander Botzki (VIB Technology Training, Flanders, Belgium) for helping with installing and hosting the project GitHub. G.T., L.P., and T.S. acknowledge the contributions of Gabriel Studer in setting up the AlphaFold pipeline for our analysis, sciCORE at the University of Basel for providing computational resources and system administration support, and funding from the SIB Swiss Institute of Bioinformatics and the Biozentrum PhD Fellowships. D.K. acknowledges supports by the National Institutes of Health (R01GM133840) and the National Science Foundation (DMS2151678, DBI2003635, CMMI1825941, MCB2146026, and MCB1925643). X.W. is recipient of the MolSSI graduate fellowship. A.M.J.J.B. and M.R. acknowledge financial support from the Netherlands eScience Center (ASDI.2016.043), from the Netherlands Organization for Scientific Research (NWO) (TOP‐PUNT grant 718.015.001) and from the European Union Horizon 2020 project BioExcel (823830). J.F.‐R. acknowledges support by Spanish Ministry of Science (grant PID2019‐110167RB‐I00 / AEI / 10.13039/501100011033). B. J.‐G. is employed by Zymvol Biomodeling on a project which received funding from the European Union's Horizon 2020 research and innovation program under Marie Skłodowska‐Curie grant agreement No. 801342 (Tecniospring INDUSTRY) and the Government of Catalonia's Agency for Business Competitiveness (ACCIÓ). X.Z. acknowledges the support from NIH/NIGMS (grant R35GM136409). E.D.L. acknowledges support from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement No. 819318), by a research grant from A.‐M. Boucher, by research grants from the Estelle Funk Foundation, the Estate of Fannie Sherr, the Estate of Albert Delighter, the Merle S. Cahn Foundation, Mrs Mildred S. Gosden, the Estate of Elizabeth Wachsman, the Arnold Bortman Family Foundation. S.D acknowledges Ramalingaswami re‐entry Fellowship (March 30th, 2021) from DBT India.O. S_F acknowledges support from the Israel Science Foundation, founded by the Israel Academy of Science and Humanities (grant number 301/2021.J.K.V. is supported by Marie Sklodowska‐Curie European Training Network Grant #860517. This study (SJW coordinator) was performed as part of the Implementation Plan (2019-2021) of Activity-2 of the 3DBioInfo Elixir Community (https://elixir-europe.org/communities/3d-bioinfo). We thank Alexander Botzki (VIB Technology Training, Flanders, Belgium) for helping with installing and hosting the project GitHub. G.T., L.P., and T.S. acknowledge the contributions of Gabriel Studer in setting up the AlphaFold pipeline for our analysis, sciCORE at the University of Basel for providing computational resources and system administration support, and funding from the SIB Swiss Institute of Bioinformatics and the Biozentrum PhD Fellowships. D.K. acknowledges supports by the National Institutes of Health (R01GM133840) and the National Science Foundation (DMS2151678, DBI2003635, CMMI1825941, MCB2146026, and MCB1925643). X.W. is recipient of the MolSSI graduate fellowship. A.M.J.J.B. and M.R. acknowledge financial support from the Netherlands eScience Center (ASDI.2016.043), from the Netherlands Organization for Scientific Research (NWO) (TOP-PUNT grant 718.015.001) and from the European Union Horizon 2020 project BioExcel (823830). J.F.-R. acknowledges support by Spanish Ministry of Science (grant PID2019-110167RB-I00 / AEI / 10.13039/501100011033). B. J.-G. is employed by Zymvol Biomodeling on a project which received funding from the European Union's Horizon 2020 research and innovation program under Marie Skłodowska-Curie grant agreement No. 801342 (Tecniospring INDUSTRY) and the Government of Catalonia's Agency for Business Competitiveness (ACCIÓ). X.Z. acknowledges the support from NIH/NIGMS (grant R35GM136409). E.D.L. acknowledges support from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement No. 819318), by a research grant from A.-M. Boucher, by research grants from the Estelle Funk Foundation, the Estate of Fannie Sherr, the Estate of Albert Delighter, the Merle S. Cahn Foundation, Mrs Mildred S. Gosden, the Estate of Elizabeth Wachsman, the Arnold Bortman Family Foundation. S.D acknowledges Ramalingaswami re-entry Fellowship (March 30th, 2021) from DBT India.O. S_F acknowledges support from the Israel Science Foundation, founded by the Israel Academy of Science and Humanities (grant number 301/2021.J.K.V. is supported by Marie Sklodowska-Curie European Training Network Grant #860517.

Funders	Funder number
ACCIÓ
Alexander Botzki
Arnold Bortman Family Foundation
Estate of Fannie Sherr
Estelle Funk Foundation
Government of Catalonia's Agency for Business Competitiveness
National Science Foundation	ASDI.2016.043, CMMI1825941, MCB1925643, DBI2003635, DMS2151678, MCB2146026
National Institutes of Health	R01GM133840
National Institute of General Medical Sciences	R35GM136409
Universität Basel
Merle S. Cahn Foundation
European Research Council	819318
Department of Biotechnology, Ministry of Science and Technology, India
Nederlandse Organisatie voor Wetenschappelijk Onderzoek	718.015.001
Israel Academy of Sciences and Humanities	860517, 301/2021
Israel Science Foundation
Vlaams Instituut voor Biotechnologie
Ministerio de Ciencia e Innovación	PID2019‐110167RB‐I00 / AEI / 10.13039/501100011033
Horizon 2020	801342, 823830
Swiss Institute of Bioinformatics

Keywords

crystal contacts
homodimers
potential energy
protein interactions
protein structure

Access to Document

10.1002/pmic.202200323

Proteomics - 2023 - Schweke - Discriminating physiological from non‐physiological interfaces in structures of proteinFinal published version, 3.54 MBLicence: Taverne

Cite this

Schweke, H., Xu, Q., Tauriello, G., Pantolini, L., Schwede, T., Cazals, F., Lhéritier, A., Fernandez-Recio, J., Rodríguez-Lumbreras, L. A., Schueler-Furman, O., Varga, J. K., Jiménez-García, B., Réau, M. F., Bonvin, A. M. J. J., Savojardo, C., Martelli, P. L., Casadio, R., Tubiana, J., Wolfson, H. J., ... Wodak, S. J. (2023). Discriminating physiological from non-physiological interfaces in structures of protein complexes: A community-wide study. Proteomics, 23(17), Article 2200323. https://doi.org/10.1002/pmic.202200323

@article{7a07a6163912445098295b4219b54c0a,

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

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.",

keywords = "crystal contacts, homodimers, potential energy, protein interactions, protein structure",

author = "Hugo Schweke and Qifang Xu and Gerardo Tauriello and Lorenzo Pantolini and Torsten Schwede and Fr{\'e}d{\'e}ric Cazals and Alix Lh{\'e}ritier and Juan Fernandez-Recio and Rodr{\'i}guez-Lumbreras, {Luis Angel} and Ora Schueler-Furman and Varga, {Julia K.} and Brian Jim{\'e}nez-Garc{\'i}a and R{\'e}au, {Manon F.} and Bonvin, {Alexandre M.J.J.} and Castrense Savojardo and Martelli, {Pier Luigi} and Rita Casadio and J{\'e}r{\^o}me Tubiana and Wolfson, {Haim J.} and Romina Oliva and Didier Barradas-Bautista and Tiziana Ricciardelli and Luigi Cavallo and {\v C}eslovas Venclovas and Kliment Olechnovi{\v c} and Raphael Guerois and Jessica Andreani and Juliette Martin and Xiao Wang and Genki Terashi and Daipayan Sarkar and Charles Christoffer and Tunde Aderinwale and Jacob Verburgt and Daisuke Kihara and Anthony Marchand and Correia, {Bruno E.} and Rui Duan and Liming Qiu and Xianjin Xu and Shuang Zhang and Xiaoqin Zou and Sucharita Dey and Dunbrack, {Roland L.} and Levy, {Emmanuel D.} and Wodak, {Shoshana J.}",

note = "Publisher Copyright: {\textcopyright} 2023 Wiley-VCH GmbH.",

year = "2023",

month = sep,

doi = "10.1002/pmic.202200323",

language = "English",

volume = "23",

journal = "Proteomics",

issn = "1615-9853",

publisher = "Wiley-VCH Verlag",

number = "17",

}

Schweke, H, Xu, Q, Tauriello, G, Pantolini, L, Schwede, T, Cazals, F, Lhéritier, A, Fernandez-Recio, J, Rodríguez-Lumbreras, LA, Schueler-Furman, O, Varga, JK, Jiménez-García, B, Réau, MF, Bonvin, AMJJ, Savojardo, C, Martelli, PL, Casadio, R, Tubiana, J, Wolfson, HJ, Oliva, R, Barradas-Bautista, D, Ricciardelli, T, Cavallo, L, Venclovas, Č, Olechnovič, K, Guerois, R, Andreani, J, Martin, J, Wang, X, Terashi, G, Sarkar, D, Christoffer, C, Aderinwale, T, Verburgt, J, Kihara, D, Marchand, A, Correia, BE, Duan, R, Qiu, L, Xu, X, Zhang, S, Zou, X, Dey, S, Dunbrack, RL, Levy, ED & Wodak, SJ 2023, 'Discriminating physiological from non-physiological interfaces in structures of protein complexes: A community-wide study', Proteomics, vol. 23, no. 17, 2200323. https://doi.org/10.1002/pmic.202200323

TY - JOUR

T1 - Discriminating physiological from non-physiological interfaces in structures of protein complexes

T2 - A community-wide study

AU - Schweke, Hugo

AU - Xu, Qifang

AU - Tauriello, Gerardo

AU - Pantolini, Lorenzo

AU - Schwede, Torsten

AU - Cazals, Frédéric

AU - Lhéritier, Alix

AU - Fernandez-Recio, Juan

AU - Rodríguez-Lumbreras, Luis Angel

AU - Schueler-Furman, Ora

AU - Varga, Julia K.

AU - Jiménez-García, Brian

AU - Réau, Manon F.

AU - Bonvin, Alexandre M.J.J.

AU - Savojardo, Castrense

AU - Martelli, Pier Luigi

AU - Casadio, Rita

AU - Tubiana, Jérôme

AU - Wolfson, Haim J.

AU - Oliva, Romina

AU - Barradas-Bautista, Didier

AU - Ricciardelli, Tiziana

AU - Cavallo, Luigi

AU - Venclovas, Česlovas

AU - Olechnovič, Kliment

AU - Guerois, Raphael

AU - Andreani, Jessica

AU - Martin, Juliette

AU - Wang, Xiao

AU - Terashi, Genki

AU - Sarkar, Daipayan

AU - Christoffer, Charles

AU - Aderinwale, Tunde

AU - Verburgt, Jacob

AU - Kihara, Daisuke

AU - Marchand, Anthony

AU - Correia, Bruno E.

AU - Duan, Rui

AU - Qiu, Liming

AU - Xu, Xianjin

AU - Zhang, Shuang

AU - Zou, Xiaoqin

AU - Dey, Sucharita

AU - Dunbrack, Roland L.

AU - Levy, Emmanuel D.

AU - Wodak, Shoshana J.

PY - 2023/9

Y1 - 2023/9

N2 - 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.

AB - 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.

KW - crystal contacts

KW - homodimers

KW - potential energy

KW - protein interactions

KW - protein structure

UR - http://www.scopus.com/inward/record.url?scp=85162927714&partnerID=8YFLogxK

U2 - 10.1002/pmic.202200323

DO - 10.1002/pmic.202200323

M3 - Article

C2 - 37365936

AN - SCOPUS:85162927714

SN - 1615-9853

VL - 23

JO - Proteomics

JF - Proteomics

IS - 17

M1 - 2200323

ER -

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

Abstract

Bibliographical note

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this