Integrating bioinformatics and chemoinformatics approaches for the development of expert systems allowing the in silico prediction of toxicities

If we could reliably predict side effects in the initial phases of drug development, this would lower the failure rate in later phases, significantly reduce the number of animal tests needed, and accelerate the development of new drugs. The eTOX project broke ground in that it enabled pharmaceutical companies to share their data on the toxicity of drug-like compounds for the first time on a large scale. This resulted in the creation of a large database, which can now be mined for further insights, including predictions on whether or not a particular compound is likely to have an adverse effect on patients. This is already helping pharmaceutical companies to make better-informed decisions in their pursuit of developing safer drugs, while at the same time reducing the use of animals in drug research.

Studies designed to predict adverse side effects in the initial phases of drug development are very important as they lower the failure rate in later phases, ensuring that companies invest resources in only the most promising drugs, and that only safe drugs reach patients. In order to improve drug safety, the pharmaceutical industry has produced vast amounts of preclinical toxicology data needed to develop compounds into drugs. After fulfilling its initial purpose, this data used to rest in individual companies’ archives, despite the fact that a lot of further knowledge could be extracted from it. However, companies used to consider this data as part of their competitive space, and so kept it to themselves.

By bringing together top experts in toxicology, knowledge management, bioinformatics, chemo-informatics, biostatistics and software development, the eTOX project set out to bring together this treasure trove of toxicological information into one database, so that it could be shared among companies for the first time, and probed for further insights. The result is a unique toxicology information database, eTOXsys, which now contains information from over 8 000 toxicity studies on almost 2 000 compounds, of which around a fifth are approved drugs. The studies done on these compounds amount to about 9 million data points about the toxicology of those compounds. This database is now commercially accessible and can be searched and mined, thanks to the tools that have been built by the academic part of the consortium.

More than 200 predictive computer-based models

Building on this treasure trove of toxicology data, which has been organised and standardised, the consortium further constructed more than 200 predictive computer-based models that have the ability to give researchers an idea of the potential toxicity of future drug-like compounds. These models make their predictions by using the complex relationships between the structure of a substance, its metabolism and disposition, and its toxic effects in the body. They enable researchers to better predict potential side effects that would otherwise only be discovered in a later stage of the drug development or when the drug is already on the market.

These computer models have been annotated in terms of the data and methods used for developing them. The academic partners in the project also developed eTOXlab, a platform for the development and maintenance of these models.

Developing a standardised language

One of the obstacles to extracting further insights from this wealth of toxicology data was that there was no common terminology between different researchers and companies. In order for all this toxicological data to be useful for further analysis, eTOX had to construct standardised terminology (ontologies). As part of this process, project researchers looked over more than 20 million entry lines, reducing more than 80 000 terms used in the initial data sources to under 7 000 preferred terms. These ontologies are now publicly available through the OntoBrowser tool and will be a valuable asset for future research.

Reducing the use of animals in research

The database and models developed within eTOX have also contributed to reducing the use of animals in research. For example, until now, pharmaceutical companies working on similar targets used to synthesise competitor compounds in order to perform laboratory studies to compare the toxicity of competitors’ compounds with their in-house candidates. Since their competitors’ toxicological data can now be found in the eTOXsys database, they no longer need to do this. According to project coordinators, there are already multiple examples of cases where information provided by eTOXsys prevented companies from running new studies on animals.

For the benefit of industry, academia, SMEs

The industry is already benefitting from the tools developed in this project, especially the eTOXsys database. For example, if a company is developing a new compound with a toxicity that is disturbing, their research teams can now go to the database to try to find a similar compound with a similar structure or a similar effect in a similar species. This is already saving time and resources, and allowing drug makers to make better-informed decisions, which could ultimately lead to safer drugs for patients. Thanks to eTOX, the industry also gained access to academic knowledge and skills, which they don’t have internally.

The academic partners have in turn benefited from learning to work together with the pharmaceutical companies, and having access to toxicity data from the industry that they otherwise would not have. The project also enabled them to develop new computer models for predicting the toxicity of drugs, as well as tools and technologies for sharing all this data.

The SMEs in the project benefitted from working with their main customers in the pharmaceutical industry, and getting to know more about who their customers are, what they want, how they think and what they do. Additionally, SMEs have gained a new business as a result of this project: eTOXsys is now commercially available and two SMEs are in charge of maintaining and managing it.

What’s next?

eTOX gave rise to a successor IMI project, eTRANSAFE, which will assess the validity of the predictive value of animal data for human safety. This successor project has the potential to modify the way researchers conduct preclinical studies, and could potentially have an even bigger impact than eTOX.

Read the interview with project coordinators

Source: IMI Innovative Medicines Initiative