ESR10: Fragment evolution platform – molecular simulations

Host: University of Barcelona, Spain
Academic supervisor: Prof. Xavier Barril (University of Barcelona)
Researcher: Maciej Majewski

Download the full description of this project: ESR10: Fragment evolution platform – molecular simulations

Hydrogen Bonds as determinants of structural stability: implications for ligand design

The aim of the project is to develop new computational approaches for structure-based drug  design. These tools will also be used to investigate new properties of macromolecular complexes and their role in molecular recognition. 

Our methods rely on a fundamental property of protein ligand complexes that has been neglected in drug design so far: structural stability, which describes the resistance of the system to structural changes. Thanks to sharp distance and angular dependencies1, stability can be provided by hydrogen bonds. Certain hydrogen bonds present strong opposition to small structural distortions and can act as kinetic traps2, thus influencing the whole dissociation process. This concept has been implemented in the molecular-dynamics based method, Dynamic Undocking (DUck)3, which allows to assess the structural stability of the complex by calculating the work used in the process of breaking the hydrogen bond (WQB).

In the first part of the project we performed a first large-scale assessment of robustness of hydrogen bonds on a set of 77 protein-ligand complexes (341 hydrogen bonds)4 sourced from the Iridium data set5. We have shown that hydrogen bond-driven structural stability is very common. Stable bonds can be found in 75% of complexes and tend to group in fragment-sized structural anchors. Additional calculations have shown that we can modulate the stability of the bond by modifying the structure of ligand. Manipulating the local environment around the bond has important implications for structural stability, and is a useful drug design principle.

In the second part of the project we used the same data set to evaluate the usefulness of structural stability in binding mode prediction. Post-docking pose evaluation with DUck was performed on a set of binding modes generated with rDock6. The results show that DUck is equally good as rDock at selecting poses.  Additionally, the performance of DUck surpasses the docking software in predicting the binding mode of the structural anchor. That has been confirmed on the set of protein-fragment complexes, gathered in SERAPhiC  data set7. DUck is also more resistant to conformational changes in the receptor, which was confirmed in cross-docking experiment. 

The project branched into a few subprojects, which were performed by other members of the group. One of them is attempting to find a relationship between structural stability and the binding free energy on the set of carefully selected activity cliffs. The other uses a large collection of DUck simulation data and machine learning approaches to construct a predictor of structural stability of macromolecular complexes.


  1. C. Bissantz, et al.J. Med. Chem. 2010,53, 5061-5084.
  2. P. Schmidtke, et al.J. Am. Chem. Soc.2011, 133, 18903-18910.
  3. S. Ruiz-Carmona, et al.Nature Chemistry2017, 9, 201.
  4. M. Majewski, et al., bioRxiv, 2018, 454165.
  5. G. L. Warren, et al.Drug Discovery Today, 2012,17, 1270-1281.
  6. S. Ruiz-Carmona, et al. PLoS computational biology, 2014, 10.4, e1003571.
  7. A. D. Favia, et al., J. Chem. Inf. Model., 2011, 51.11, 2882-2896.

Publications and Dissemination


  • Maciej Majewski, and Xavier Barril. ” Structural Stability Predicts the Binding Mode of Protein-Ligand Complexes” manuscript in preparation
  • Maciej Majewski, Sergio Ruiz-Carmona, and Xavier Barril. “An investigation of structural stability in protein-ligand complexes reveals the balance between order and disorder.” Communications Chemistry 2.1 (2019): 1-8.
  • Maciej Majewski, Sergio Ruiz-Carmona, and Xavier Barril. “Dynamic Undocking: A Novel Method for Structure-Based Drug Discovery.” Rational Drug Design. Humana Press, New York, NY, 2018. 195-215.

Oral presentations:

  • 11th International Conference on Chemical Structures, 27-31/5/2018, Noordwijkerhout, The Netherlands, Hydrogen Bonds as Determinants of Structural Stability, M. Majewski, S. Ruiz-Carmona, X. Barril
  • Gordon Research Seminar 2017 – Computer Aided Drug Design, 15-16.07.2017, Mount Snow, VT, USA, Extending the Quasi-Bound concept to multiple interaction points, M. Majewski, S. Ruiz-Carmona, X. Barril


  • Gordon Research Conference 2019 – Computer Aided Drug Design, 14-19.07.2019, Mount Snow, VT, USA, Structural Stability in Protein-Ligand Complexes: Implications for drug design, M. Majewski, S. Ruiz-Carmona, X. Barril
  • Fragments 2019: 7th RSC-BMCS Fragment-based Drug Discovery meeting, 24-26.03.2019, Cambridge, UK, The assessment of structural stability of hydrogen bonds on the complexes with fragments and drug-like molecules, M. Majewski, S. Ruiz-Carmona, X. Barril
  • ISQBP President’s Meeting 2018, 16-21.06.2018, Barcelona, Spain, Application of Dynamic Undocking in Binding Mode prediction, M. Majewski, S. Ruiz-Carmona, X. Barril
  • Gordon Research Conference 2017 – Computer Aided Drug Design, 16-21.07.2017, Mount Snow, VT, USA, Extending the Quasi-Bound concept to multiple interaction points, M. Majewski, S. Ruiz-Carmona, X. Barril

Other projects

Contact details

Please contact us at:

FRAGNET Coordinator
VU University Amsterdam
The Netherlands

Funded by

Marie Curie Actions

EU Horizon 2020

European Union