Successful modeling of a compound in a biological system
Clean compound and bioactivity data are essential to successful modeling of the impact of a compound on a biological system
e-Therapeutics is a drug development group in Oxford, UK, that uses a network biology approach to discover novel drug candidate for biologically complex diseases. Trained as a computational neuroscientist, Dr. Johnny Wray joined e-Therapeutics in hopes of solving a very real problem in pharmaceutical development by using network models.
In building and observing its network models, e-Therapeutics relies on its virtual compound library as a vital component of its workflow. The library contains roughly 10 million compounds — approximately half with known bioactivity data and the other half with activity information predicted by machine learning.
The hardest part in building our compound library was establishing the pipeline to extract, transform and load data from multiple sources, each with its own structure. Our biggest issue is cleaning up source data and this is no simple task. Things as simple as spelling mistakes can impair our pipeline, Wray explained.
The data from Reaxys is the cleanest we have access to. The effort put into standardizing and normalizing Reaxys makes them easy to map to our internal database structure, said Wray. But there was another advantage that was unexpected.
What surprised me in the implementation of data integration from all our sources is that there is really minimal overlap in content. This means that the contribution of Reaxys is quite substantial and my overall impression is that Reaxys excels not only in the number of compounds, but also in terms of the comprehensiveness of bioactivity and target information.
We have empirically proven that our network-based approach to drug discovery works.
Dr. Johnny Wray
Today, there seems to be a paradigm shift in the methodologies being used to identify and optimize new drugs.
We know that compounds often affect more than one target and that future drug development must accommodate and even capitalize on that. Phenotypic screening is also becoming more popular again. Independent of the idea of networks, the growing appreciation that no one gene can be responsible for complex phenotypes, and that it is necessary to measure multiple factors to truly assess the value of a compound as a drug, is heartening, Wray explained.
It means we are moving in the right direction!
Download the full case study to learn more (PDF, 669.8 KB)
Elsevier R&D Solutions supporting e-Therapeutics
Enable better informed preclinical drug safety decisions with comprehensive, fully searchable FDA/EMA drug approval documents and comparative extracted drug safety data.