X-ray structure of the Epidermal Growth Factor Receptor (EGFR) with gefitinib bound (PDB id: 4WKQ)

Rational drug design refers to designing drug molecules that bind to a target (e.g. protein, nucleic acid). It relies on prior knowledge of the structure, function, and mechanism of the target, thereby avoiding random testing of thousands of molecules. Computer-aided drug design is crucial in supporting medicinal chemists in the target-to-hit, hit-to-lead, and lead optimization phases towards a drug candidate. Among these methods, we focus here on structure-based drug design and molecular dynamics-based approaches. The software packages used here are BioExcel Building Blocks library (biobb), GROMACS, PyCOMPSs, HADDOCK, autodock, pmx, CWL, PMUT, OpenBabel, and ACPype.

Why is this project particularly interesting for BioExcel?

In the early days of drug development, the optimization of chemical series was largely a trial-and-error process guided by the intuition of medicinal chemists. Over the last 50 years, rational drug design has been widely used as a more knowledge-based alternative. It builds upon the structure of the target as well as biological activity and property data of chemical series. Compared to high throughput screening alone, it cuts costs and time expenditures for pharmaceutical companies considerably.

There are however two main bottlenecks to the routine use of rational drug design techniques in the pharmaceutical industry: 1) Existing computational pipelines typically use many different software for a variety of tasks (modelling, simulation, docking, etc.) with a clear lack of interoperability, and 2) the computational cost involved in the drug design process.

What are we doing in BioExcel? 

The BioExcel Centre of Excellence aims at tackling these bottlenecks by providing interoperable computational biomolecular building blocks that were developed in the first phase of the project, by creating a set of rational drug design workflows, and by connecting these to large scale high-performance computing (HPC) resources. These workflows streamline common computer-aided drug design tasks and allow delivery of  results in a timely fashion.

BioExcel’s rational drug design project unites several high impact computer-aided drug design techniques in biomolecular workflows that are relevant to the pharmaceutical industry. We have divided the project into four different studies of increasing complexity. Each study will have its dedicated workflow (WF).

  •   WF1: Moving mutational analysis into 
the structural field for drug design
  •   WF2: Tackling mutations inactivating tumor suppressors
  •   WF3: Quantitative predictions of binding affinity in lead optimization
  •   WF4: Machine learning for efficient drug design

The usability of the WFs will be assessed by Nostrum Biodiscovery: Systems with relevance to the pharmaceutical industry will serve as validations. The workflow prototypes will be tested with massively parallel executions on the infrastructures of the Barcelona Supercomputing Center.  


IRB Barcelona, Barcelona Supercomputing Center, Nostrum Biodiscovery, Max Planck Gesellschaft, KTH Royal Institute of Technology, The University of Manchester