Ferran, Sanz, Pognan, Francois, Steger-Hartmann, Thomas, Díaz, Carlos, Montserrat, Cases, Manuel, Pastor, Marc, Philippe, Wichard, Joerg, Briggs, Katherine, Watson, David, Schwab, Christof H. , Yang, Chihae, Amberg, Alexander , Beaumont, Maria, Brookes, Anthony, Brunak, Søren, Cronin, Mark T. D., Ecker, Gerhard F. , Escher, Sylvia, Greene, Nigel, Guzmán, Antonio, Hersey, Anne, Jacques, Pascale, Lammens, Lieve, Jordi, Mestres, Muster, Wolfgang, Northeved, Helle, Pinches, Marc, Sáiz, Javier, Sajot, Nicolas, Valencia, Alfonso, Van Der Lei, Johan , Vermeulen, Nico P. E., Vock, Esther, Wolber, Gerhard and Zamora, Ismael (2017) Legacy data sharing to improve drug safety assessment: the eTOX project. Nature Reviews Drug Discovery, 16 (12). pp. 811-812. ISSN 1474-17761474-1784

Abstract

The proposed manuscript would start with a justification of the relevance of data sharing (1) for improving the in silico prediction of organ toxicities and a brief description of the main characteristics of the eTOX IMI project (2010-2016) (http://www.etoxproject.eu) (2), followed by an explanation of its main achievements and the challenges faced and lessons learned during its execution. The paper will also incorporate some examples of how the pharmaceutical companies are using the system that has been developed.
We would explain that the main goals of eTOX were the building of a unique database constituted by toxicological reports donated by the 13 pharmaceutical companies that participated in the project, and the use of the database and other information sources for enabling more effective read-across and predictive modeling to be applied in early stages of the drug development. The most recent version of the eTOX database contains 8,196 toxicological studies done on 1,947 compounds, which include almost nine million findings (clinical, histopathological, from gross necropsy, etc.). On the other hand, the eTOX partners have developed more than 200 predictive models, in which innovative strategies have been applied (3-5). Furthermore, an integrated and user-friendly software platform (eTOXsys) has been set up to ease the exploration of the database and the use of the models. The system incorporates an exploratory module that queries possible human safety liabilities for a compound on the basis of its similarity to marketed drugs.
Regarding the challenges and lessons, the first one was related to the resistances that we had to be overcome for making possible that pharmaceutical companies used to compete among them, accept to share valuable proprietary pre-clinical data. This required a combination of technical, psychological, legal, political, social (snowball) and business solutions. A second challenge was the lack of standardization of the data contributed by each company. This implied the cooperative development and implementation of relevant ontologies (6). Curation campaigns were also required to increase the data quality. On the other hand, the extraction of relevant information from free text implied the application of text mining techniques (7). Moreover, the raw data included in the database was not directly suitable for modeling and had to be transformed into adequate variables. The strict documentation and versioning of the models, as well as the assessment of their performance and applicability domain was another requirement to satisfy. The procedures that were implemented and the tools that were developed for such a purpose (8,9) will be discussed.
Finally, we would comment the post grant state of eTOX and we would briefly introduce recently granted IMI projects (TransQST and eTRANSAFE) that will supplement the work done in eTOX.

Item Type:	Article
Keywords:	IMI, eTOX, consortium, data sharing, FAIR, ontologies, visualisation
Date Deposited:	12 Dec 2017 00:45
Last Modified:	12 Dec 2017 00:45
URI:	https://oak.novartis.com/id/eprint/33611