Project 2. Pharmacovigilance-guided drug design.


Key project members: Simon Folkertsma, Ganesh Vaidyanathan.


Sponsor: Biorange and Schering-Plough



Pharmacovigilance-guided drug design is about the development of new tools and scientific approaches to identify correlations between a safety and efficacy profile of a drug in human (based on postmarket safety data reported by consumers, investigator etc.) and the properties of a drug (e.g. structural or omics derived data).  Basically, the project is strongly connected to the technologies and key goals of the Biorange Project 1 ‘Exploiting Structural Genomics Information to Incorporate Protein Flexibility in Drug Design' in sharing the aim to identify high quality and more drug-like leads. This to improve success in lead optimization and to reduce high attrition rates in the clinical phase of the drug development process.

One of the reasons for late stage attrition is the lack of sufficient and/or inconclusive predictive models in Research. To reduce these costly late-stage failures and to direct basic discoveries into the clinic more effectively it is crucial to enhance predictive capabilities by creating an optimal alignment between discovery research/exploratory development and full development. The exchange of ideas, ways of thinking, tools, and information between Research and Development is critical.

From this perspective, we want to explore the use of postmarketed Pharmacovigilance data in Research. Pharmacovigilance is the science and activities relating to the detection, assessment, understanding and prevention of adverse effects or any other drug-related problems. Pharmacovigilance is important in running clinical trials to detect emergent safety issues of the drug in development. Post market, Pharmacovigilance plays an important role in the collection of adverse effects of marketed drugs, which have led to a growing volume of high quality safety information available in (public) databases in recent years.

The goal of the project is to design an information platform for Pharmacovigilance-guided drug design and to implement translational informatics approaches to link clinical safety signals in this post market data with the structural properties or induced gene expression profiles of drug molecules. When a link is found, this information can be used in a translational way, for predicting adverse events and efficacy at the chemical design phase. Potentially, the use of Pharmacovigilance data in early Research can assist scientist in the complex decision making process (including repositioning of drugs) and enables proactive risk management


Project Goals

  • Development of an integrated drug structures database and adverse event database
  • Identify relations between Toxicogenomics profiles of (known) drugs and drug safety profiles in human, based on pharmacovigilance signaling studies.
  • Deliver systematic methods for design of drugs with a better safety profile with the ability to identify correlations between structural properties of compounds (and targets) and human safety data (use of human safety data in research)
  • New integrated in silico approaches to reduce late stage attrition in drug discovery & development
  • Prediction of human toxicity at early phase in Research
  • Identification of genes underlying human adverse events (biomarkers)



This project is a joint industry-academic initiative. Within Schering-Plough, state-of-the-art commercial safety signaling software is available that creates safety profiles based on postmarket data by use of various algorithms. This software will be used within the project to study the correlation between a drug's safety profile and its structure or induced gene expression profile, this to establish proof of concept (POC) of the method. To achieve POC, we will first focus on 95 compounds for which we have both high quality Toxicogenomics profiles in liver and human safety data profiles created from the post market safety data with the in-house software. Clustering of both profile sets will reveal insight in the correlation between the two datasets, ideally on the level of individual genes that underlie specific safety problems (biomarker detection).

In parallel, the academic partner will develop an information platform which facilitates the use of human safety data in Research. Key component is a relational database that stores the publicly available adverse event data of the Food and Drug Administration (FDA). Since drug naming in adverse event reports is not standardized we aim to develop cleaning software (based on text mining algorithms developed for CoPub) as an additional component of the platform. At a later stage, signaling software will be added to the platform, which can be used for the creation of safety profiles of drugs in the database.

In contrast to existing pharmacovigilance software packages, which are intended to detect signals very precisely for drug safety reasons and are used in a regulatory environment, we aim to develop a database and software that fits better in a Research environment and which can be used in various academic Translational Medicine projects.


Results to date

  • Preliminary methods set up for omics-based and structure-based pharmacovigilance.
  • Basic sql database storing the FDA Adverse Event Reporting System (AERS) adverse

event data

  • Text mining scripts to clean the drug naming inconsistencies in the FDA AERS data
  • Strategy to implement the connection between the drug names in the FDA AERS

        database and their 2D/3D structures, including ATC codes.


Innovative elements

  • Pharmacovigilance data in Translational Medicine
  • First academic approach to create a human safety database