h1

TY  - THES
AU  - Raghavan, Bharath
TI  - High performance computing-based QM/MM simulations for drug design: application to the non-invasive diagnosis of IDH1-associated glioma
PB  - RWTH Aachen University
VL  - Dissertation
CY  - Aachen
M1  - RWTH-2024-07554
SP  - 1 Online-Ressource : Illustrationen
PY  - 2024
N1  - Veröffentlicht auf dem Publikationsserver der RWTH Aachen University
N1  - Dissertation, RWTH Aachen University, 2024
AB  - Structure-based drug design and molecular dynamics (MD), are routinely used to speed up the pre-clinical stages of the drug discovery process by predicting ligand-target poses. Nevertheless, current MD methods (based on classical physics), may encounter difficulties in describing a large range of targets, including many metal-containing proteins, and enzyme transition states. This greatly limits the applicability of MD to drug discovery. A quantum mechanical (QM) treatment may be required to accurately investigate these targets. However, this is very expensive and infeasible to implement in current drug design pipelines. Here I show that the density functional theory-based multiscale Quantum Mechanical/Molecule Mechanical (QM/MM) MD, as implemented in the recently introduced MiMiC framework with high performance computing (HPC), has a great potential to contribute towards such efforts. A quantum-based HPC virtual screening (QHPC-VS) protocol, integrating MiMiC-QM/MM MD, is formulated and refined here. This includes the development of the MiMiCPy software package for seamless conversion of MM outputs to QM/MM files of large biomolecules. My protocol is used to study the therapeutically important Isocitrate Dehydrogenase 1 enzyme. Very high performance and scaling of MiMiC was achieved for this application, demonstrating its potential for QM/MM simulations of biology. The mutant (R132H) isoform of IDH1 has been noted as a potential predictive biomarker for glioma, introducing a strong rationale to develop non-invasive imaging methods of this marker like positron emission tomogrpahy (PET). This involves targeting the mutant IDH1 active site with selective PET radioactive ligands or radiotracers. However, due to the nature of the protein, such selective ligands have not been proposed in the literature so far. Furthermore, classical MD simulations of the complicated active site lead to extreme distortion of the structure, making application of computational techniques dificult. The QHPC-VS pipeline allowed for accurately simulating the mut-IDH1 active site at the quantum level, and suggest ligands that would function as radioactive tracers for PET. Our simulations predict 15 such small molecules, out of which is fluorothymidine, an already well-known PET radiotracer precursor for various other cancers. This discovery has the potential to greatly accelerate the development of non-invasive diagnosis of glioma.
LB  - PUB:(DE-HGF)11
DO  - DOI:10.18154/RWTH-2024-07554
UR  - https://publications.rwth-aachen.de/record/990890
ER  -