Ten big pharmaceutical entities, including GSK, AstraZeneca, and Johnson & Johnson, have ventured into their first partnership aimed at training their machine-learning algorithms for drug discovery using each other’s data.
Owkin, a New York and Paris-based startup that is backed by Google Ventures, has come up with a secure, blockchain powered system that enables an algorithm to sift through a competitor’s data with complete traceability – but carefully without exposing commercial secrets to business rivals.
The machine learning algorithm helps in improving the models that forecast how different molecules operate, with the hope of expediting the cumbersome and costly process of drug discovery.
According to details found in a 2016 paper published in the Journal of Health Economics, on average, the process costs about €1.9 billion and takes 13 years before the drug can be introduced into the market.
Mathieu Galtier, the coordinator of the particular project, asserted that the startup was aiming at considerably accelerating the understanding and search process of the various compounds that may work as drugs.
“The big goal here is to accelerate and reduce the cost of the discovery of drugs,” he said.
Currently, investors are investing heavily in startups that are trying to leverage AI in improving the sometimes messy drug discovery process.
In the United States, Schrodinger, a David E Shaw and Bill Gates-backed platform, recently completed a $110 million financing round.
Back in 2018, Relay Therapeutics fundraised $400 million from investors such as the SoftBank Vision Fund whereas Insitro recently raised over $100 million from venture capitalists, including Andreessen Horowitz.
In the United Kingdom, Genomics, a University of Oxford spinout, entered into a drug discovery deal with biotech company Vertex Pharmaceuticals.
Even though pharmaceutical companies are customers of such start-ups, the Melloddy project, which translates to Machine Learning Ledger Orchestration for Drug Discovery, represents the first time that such entities have shared their data amongst themselves.
If the project is successful, it could end up encouraging competitor companies to open up more of their data, for instance, the outcomes of their pre-clinical studies.
The leader of the Melloddy project Hugo Ceulemans claimed that the initiative can boost patient care.
“This project allows the pharma partners for the first time to collaborate in their core competitive space, invigorating discovery efforts through efficiency gains,” said Mr Ceulemans, who also serves as the scientific director of discovery data sciences at Janssen Pharmaceutica, a Johnson & Johnson’s drugmaker based in Belgium.
The Melloddy research division boasts about €10 million “in kind” financing from each entity involved and €8 million from the European Union Innovative Medicines Initiative.
The consortium is made up of 17 partners including Nvidia, four start-ups, 10 pharmaceutical companies, and two European universities.
By utilizing “federated learning”, a type of decentralized machine learning, pharmaceutical companies are able to store their data away from their chemical libraries.
“We want absolute traceability of all operations made on the platform. It is very important that each pharma partner knows they are being treated on equal grounds, to make sure what is happening to their data is transparent so they can check it afterwards,” he said.
The MELLODDY consortium consists of 17 partners:
- 10 pharmaceutical companies: Amgen, Astellas, AstraZeneca, Bayer, Boehringer Ingelheim, GSK, Janssen Pharmaceutica NV, Merck KgaA, Novartis, and Institut de Recherches Servier
- 2 academic universities: KU Leuven, Budapesti Muszaki es Gazdasagtudomanyi Egyetem
- 4 subject matter experts: Owkin, Substra Foundation, Loodse, Iktos
- 1 large AI computing company: NVIDIA