Industry-Scale Orchestrated Federated Learning for Drug Discovery

To apply federated learning to drug discovery we developed a novel platform in the context of European Innovative Medicines Initiative (IMI) project MELLODDY (grant n{\deg}831472), which was comprised of 10 pharmaceutical companies, academic research labs, large industrial companies and startups. Th...

Full description

Saved in:
Bibliographic Details
Published inarXiv.org
Main Authors Oldenhof, Martijn, Ács, Gergely, Pejó, Balázs, Schuffenhauer, Ansgar, Holway, Nicholas, Sturm, Noé, Dieckmann, Arne, tmeier, Oliver, Boniface, Eric, Mayer, Clément, Gohier, Arnaud, Schmidtke, Peter, Niwayama, Ritsuya, Kopecky, Dieter, Lewis, Mervin, Rathi, Prakash Chandra, Friedrich, Lukas, manek, András, Antal, Peter, Rahaman, Jordon, Zalewski, Adam, Heyndrickx, Wouter, Oluoch, Ezron, Stößel, Manuel, Vančo, Michal, Endico, David, Gelus, Fabien, Thaïs de Boisfossé, Darbier, Adrien, Nicollet, Ashley, Blottière, Matthieu, Telenczuk, Maria, Van Tien Nguyen, Martinez, Thibaud, Boillet, Camille, Moutet, Kelvin, Picosson, Alexandre, Gasser, Aurélien, Inal Djafar, Simon, Antoine, Arany, Ádám, Simm, Jaak, Moreau, Yves, Engkvist, Ola, Ceulemans, Hugo, Marini, Camille, Galtier, Mathieu
Format Paper
LanguageEnglish
Published Ithaca Cornell University Library, arXiv.org 12.12.2022
Subjects
Codification
Cryptography
Federated learning
Iterative methods
Web services
Online AccessGet full text

Cover

More Information
Summary:To apply federated learning to drug discovery we developed a novel platform in the context of European Innovative Medicines Initiative (IMI) project MELLODDY (grant n{\deg}831472), which was comprised of 10 pharmaceutical companies, academic research labs, large industrial companies and startups. The MELLODDY platform was the first industry-scale platform to enable the creation of a global federated model for drug discovery without sharing the confidential data sets of the individual partners. The federated model was trained on the platform by aggregating the gradients of all contributing partners in a cryptographic, secure way following each training iteration. The platform was deployed on an Amazon Web Services (AWS) multi-account architecture running Kubernetes clusters in private subnets. Organisationally, the roles of the different partners were codified as different rights and permissions on the platform and administrated in a decentralized way. The MELLODDY platform generated new scientific discoveries which are described in a companion paper.
ISSN:2331-8422