Skip to Main content Skip to Navigation
Theses

Hydration of drug-like molecules with molecular density functional theory and the hybrid-4th-dimension Monte Carlo approach

Abstract : The development of a drug takes on average over 10 yr. for a cost of 1B dollars. To speed up the process, and reduce its cost, in-silico methods are used at the drug discovery stage. It consists of screening ~10⁵ drug-like molecules to propose few candidates to the pre-clinical stages. The main criterion is the affinity between the potential drug molecule and biological target. As the interaction happens the body, these affinities need to be predicted in water and the molecule needs to be water-soluble to access the receptor. Overall, solvation properties play an important role in drug design. Numerically, for a given force-field, solvation can be studied either with exact but time-consuming simulation methods, fast continuum models that lose the molecular nature of the solvent, or approximate liquid state theories that keep the solvent molecular information while speeding-up the computation. In this thesis, we focus on the prediction of the hydration free energies (HFE) of drug-like molecules with methods that are as fast and precise as possible, and we concentrate on two original approaches: Hybrid-4th-dimension Monte Carlo, a novel method that computes the HFEs according to the Jarzynski principle from short non-equilibrium simulations in which the solute is inserted or removed from the solvent with a time-depending coupling parameter. This approach is shown to predict the HFEs of drug-like molecules 4-6 times faster than the classical free energy perturbation approach. Molecular density functional theory, a liquid-state-theory approach that allows the study of the equilibrium solvation properties of any rigid solute. In its current level, the hyper netted-chain approximation coupled with a pressure correction, it is shown to predict the HFEs of drug-like molecules within 0.5 and 1.0 kcal/mol of simulations and experimental data, respectively, for an average computational speed-up 10³-10⁴ with respect to simulations. H4D-MC is considered here as a source of reference data for MDFT developments. MDFT is itself fast enough to be foreseen in a high-throughput screening pipeline.
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-03121661
Contributor : Abes Star :  Contact
Submitted on : Tuesday, January 26, 2021 - 3:21:15 PM
Last modification on : Thursday, January 28, 2021 - 3:29:01 AM

File

97404_LUUKKONEN_2020_archivage...
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-03121661, version 1

Collections

Citation

Sohvi Luukkonen. Hydration of drug-like molecules with molecular density functional theory and the hybrid-4th-dimension Monte Carlo approach. Theoretical and/or physical chemistry. Université Paris-Saclay, 2020. English. ⟨NNT : 2020UPASF030⟩. ⟨tel-03121661⟩

Share

Metrics

Record views

59

Files downloads

35