Posts archive
-
PhD fellowship in collaboration with IMEC (Leuven) and our research group (UA)
We are seeking a highly motivated and talented PhD candidate to join our collaborative research project between the AI & Algorithms group at IMEC (Leuven) and the University of Antwerp’s Laboratory of Medicinal Chemistry (UAMC), under the guidance of Prof. Hans De Winter. This is a unique opportunity to work at the intersection of cutting-edge technology and medicinal chemistry, contributing to the development of novel inhibitors for therapeutic protein targets through computational methods.
-
Cosolvent MD on DPP4, DPP8 and DPP9
In this recent Journal of Chemical Information and Modeling paper by Olivier Beyens, our efforts in computational drug design against dipeptidyl peptidase 4 (DPP4), DPP8 and DPP9 are presented. Cosolvent molecular dynamics (MD) simulations have been applied to these three protein targets. The cosolvent molecular dynamics simulations reproduce key ligand binding features and known binding pockets, while also highlighting interesting fragment positions for future ligand optimization. The resulting fragment maps from the cosolvent molecular dynamics are freely available for use in future research (). Detailed instructions for easy visualization of the fragment maps are provided, ensuring that the results are usable by both computational and medicinal chemists. Additionally, we used the fragment maps to search for the binding pockets with significant potential using an algorithmic approach combining top fragment locations. To discover novel binding scaffolds, a limited pharmacophore screening was performed, where the pharmacophores were based on the analyses of the cosolvent simulations.
-
Results from our collaboration with reMYND published in Science
On May 31, 2024, the results from our collaboration with scientists from the biotechnology company reMYND have been published in Science. In this collaboration, we used extensive LiGaMD simulations to confirm the binding location of reMYND’s lead compound within a complex of several septin proteins. We also performed umbrella sampling to qualitatively estimate the impact of this binding on the stability of the septin-6/septin-7 heterodimer. This interdisciplinary research work is a perfect example of the use of computational tools to validate and explain biological hypotheses. Many thanks to Marc Fivaz and all his colleagues of reMYND!
-
PhD defence of Alan Kerstjens
On Tuesday April 26, 2024, our student Alan Kerstjens succesfully defended his PhD thesis to become Doctor in the Pharmaceutical Sciences. His work was entitled “Computational design of synthesizable molecules by imitating reference chemistry”. This work was build around several of the software packages that Alan wrote, including LEADD, MolPert and MoleculeAutoCorrect. External jurymembers were Greg Landrum (ETH Zurich) and Mazen Ahmad (J&J). Many thanks and congrats to Alan!
-
Preventing lipophilic aggregation in cosolvent molecular dynamics simulations with hydrophobic probes using Plumed Automatic Restraining Tool (PART)
Cosolvent molecular dynamics simulations are molecular dynamics simulations used to identify preferable locations of small organic fragments on a protein target. Most cosolvent molecular dynamics workflows make use of only water-soluble fragments, as hydrophobic fragments would cause lipophilic aggregation. To date the two approaches that allow usage of hydrophobic cosolvent molecules are to use a low concentration of hydrophobic probes, with the disadvantage of a lower sampling speed, or to use force field modifications, with the disadvantage of a difficult and inflexible setup procedure. In this new paper [J. Cheminformatics (2024), 16, 23] we present a new alternative, that does not suffer from low sampling speed nor from cumbersome preparation procedures. We have built an easy-to-use open source command line tool PART (Plumed Automatic Restraining Tool) to generate a PLUMED file handling all intermolecular restraints to prevent lipophilic aggregation. We have compared restrained and unrestrained cosolvent MD simulations, showing that restraints are necessary to prevent lipophilic aggregation at hydrophobic probe concentrations of 0.5 M. Furthermore, we benchmarked PART generated restraints on a test set of four proteins (Factor-Xa, HIV protease, P38 MAP kinase and RNase A), showing that cosolvent MD with PART generated restraints qualitatively reproduces binding features of cocrystallised ligands.
-
Molecule auto‑correction to facilitate molecular design
Ensuring that computationally designed molecules are chemically reasonable is at best cumbersome. In this new paper [J. Comput. Aided Mol. Des. (2024), 38, 10] by Alan Kerstjens we present a molecule correction algorithm that morphs invalid molecular graphs into structurally related valid analogs. The algorithm is implemented as a tree search, guided by a set of policies to minimize its cost. We showcase how the algorithm can be applied to molecular design, either as a post-processing step or as an integral part of molecule generators.
-
A molecule perturbation software library and its application to study the effects of molecular design constraints
“A molecule perturbation software library and its application to study the effects of molecular design constraints” is our recent paper [J. Cheminform. (2023), 15, 89] that has just been accepted in Journal of Cheminformatics. In this work we present a software library for constrained graph-based molecule manipulation and showcase its functionality by developing a molecule generator. Such a generator designs molecules mimicking reference chemical features of differing granularity. We find that restricting molecular construction lightly, beyond the usual positive effects on drug-likeness and synthesizability of designed molecules, provides guidance to optimization algorithms navigating chemical space.
-
Computational approaches streamlining drug discovery
A splendid review on computer-aided drug discovery was recently published in Nature. In this paper, the authors describe the rise of computational drug design in recent years due to the abundance of data on ligand properties, computing power, and virtual libraries of drug-like molecules. Computational drug design has the potential to reshape the drug discovery and development process, but it also faces challenges. At Silicos-it, we are equipped to help you with these tasks, so please do not hesitate to contact us if you have a project that could benefit from these novel approaches.
-
Interesting paper on AlphaFold2
Researcher at the University of Leipzig and Vanderbilt University have published an interesting paper [Front. Mol. Biosci. (2023), 10, 1121962] that describes a new AlphaFold pipeline to model GPCRs and kinases with user-defined conformational states.
-
EuroHPC computing time granted
We just received a computing grant on the Tier-1 supercomputer Karolina. The available GPU computing resources will be used to benchmark some LiGaMD-based molecular dynamics simulations using the AMBER20 software tools.
-
LUMI computing time granted
We just received a computing grant on the new exascale Tier-0 supercomputer LUMI. The available GPU computing resources will be used to study the dimerisation energetics of DPP9, and how this is influenced by inhibitor binding. State-of-the-art Gaussian-accelerated molecular dynamics simulations will be applied.
-
VLAIO grant MOSA approved
We have just received a VLAIO grant for a project on multi-objective compound design using state-of-the-art AI models (MOSA). This work is a joint collaboration with Johnson & Johnson, the group of Gianni de Fabritiis in Barcelona, and Open Analytics in Belgium.
-
LEADD: Lamarckian evolutionary algorithm for de novo drug design
“LEADD: Lamarckian evolutionary algorithm for de novo drug design” is our recent paper that has just been accepted [J. Cheminform. (2022), 14, 3]. In this paper we describe a Lamarckian evolutionary algorithm for de novo drug design (LEADD).
-
ESF-project E-learning approved
We have just received a grant of the ESF to design and build a virtual reality platform to visualise protein-ligand interactions. In addition, the project will be used to create a blended learning environment for students to learn computational drug design as part of the Biochemistry curriculum at the University of Antwerp. This project is in close collaboration with Soulmade who implemented the VR application. The resulting blended-learning curriculum is available at Github:
-
Our Shape-it software tool has been incorporated into the RDKit suite
Thanks to the excellent work of Greg Landrum and Iwatobipen, our Shape-it screening tool has now been ported into the RDKit suite of chemo-informatics tools.
-
VLAIO grant CryoStudio approved
We have just received a VLAIO grant for a project focussed on structural analysis of cryo-EM models and their use in lead optimisation (CryoStudio). This work is a joint collaboration with Johnson & Johnson and the group of Erik Lindahl in Sweden.
-
Assessment of the fragment docking program SEED
“Assessment of the Fragment Docking Program SEED” is our recent paper that has just been accepted [J. Chem. Inf. Model. (2020), 60, 4881-4893]. In this paper we evaluate the performance of the fragment docking program SEED (Solvation Energy for Exhaustive Docking) on 15 different protein targets.
-
A QM/MM study on the reaction mechanism of Staphylococcus Aureus monoglycosyltransferase
Our recent paper has just been accepted in Journal of Chemical Information and Modeling [J. Chem. Inf. Model. (2020), 60, 5513-5528]. In this paper we investigate the catalytic mechanism of a bacterial monoglycosyltransferase enzyme using QM/MM simulations. From these results, we present a new hypothesis for the binding mode of lipid II and the reaction mechanism of the GT51 glycosyltransferases.
-
Molecular dynamics simulations of membrane proteins: an overview
Our recent review has just been accepted in Journal of Chemical Information and Modeling [J. Chem. Inf. Model. (2018), 58, 2193-2202]. In this review we provide an overview of the history and current state-of-the-art methodologies in membrane protein simulations.
-
Virtual screening for inhibitors of the human TSLP:TSLPR interaction
Our recent work on the virtual screening of small molecules to inhibit the TSLP:TSLPR interaction has just been published in Scientific Reports [Sci. Rep. (2017), 7, 17211]. This work provides a proof-of-principle for using fragments in the inhibition of TSLP:TSLPR complexation.