Development of new products often relies on the discovery of novel molecules.While conventional molecular design involves using human expertise to propose, synthesize, and test new molecules, this process can be cost and time intensive, limiting the number of molecules that can be reasonably tested. Generative modeling provides an alternative approach to molecular discovery by reformulating molecular design as an inverse design problem. Here, we review the recent advances in the state-of-the-art of generative molecular design and discusses the considerations for integrating these models into real molecular discovery campaigns. We first review the model design choices required to develop and train a generative model including common 1D, 2D, and 3D representations of molecules and typical generative modeling neural network architectures. We then describe different problem statements for molecular discovery applications and explore the benchmarks used to evaluate models based on those problem statements. Finally, we discuss the important factors that play a role in integrating generative models into experimental workflows. Our aim is that this review will equip the reader with the information and context necessary to utilize generative modeling within their domain.
Accurately predicting small molecule partitioning and hydrophobicity is critical in the drug discovery process. There are many heterogeneous chemical environments within a cell and entire human body. For example, drugs must be able to cross the hydrophobic cellular membrane to reach their intracellular targets, and hydrophobicity is an important driving force for drug−protein binding. Atomistic molecular dynamics (MD) simulations are routinely used to calculate free energies of small molecules binding to proteins, crossing lipid membranes, and solvation but are computationally expensive. Machine learning (ML) and empirical methods are also used throughout drug discovery but rely on experimental data, limiting the domain of applicability. We present atomistic MD simulations calculating 15,000 small molecule free energies of transfer from water to cyclohexane. This large data set is used to train ML models that predict the free energies of transfer. We show that a spatial graph neural network model achieves the highest accuracy, followed closely by a 3D-convolutional neural network, and shallow learning based on the chemical fingerprint is significantly less accurate. A mean absolute error of ∼4 kJ/mol compared to the MD calculations was achieved for our best ML model. We also show that including data from the MD simulation improves the predictions, tests the transferability of each model to a diverse set of molecules, and show multitask learning improves the predictions. This work provides insight into the hydrophobicity of small molecules and ML cheminformatics modeling, and our data set will be useful for designing and testing future ML cheminformatics methods.
Simulations and experiments have revealed enormous transport rates through carbon nanotube (CNT) channels when a pressure gradient drives fluid flow, but comparatively little attention has been given to concentration‐driven transport despite its importance in many fields. Here, membranes are fabricated with a known number of single‐walled CNTs as fluid transport pathways to precisely quantify the diffusive flow through CNTs. Contrary to early experimental studies that assumed bulk or hindered diffusion, measurements in this work indicate that the permeability of small ions through single‐walled CNT channels is more than an order of magnitude higher than through the bulk. This flow enhancement scales with the ion free energy of transfer from bulk solutions to a nanoconfined, lower‐dielectric environment. Reported results suggest that CNT membranes can unlock dialysis processes with unprecedented efficiency.
Understanding ion solvation and transport under confinement is critical for a wide range of emerging technologies, including water desalination and energy storage. While molecular dynamics (MD) simulations have been widely used to study the behavior of confined ions, considerable deviations between simulation results depending on the specific treatment of intermolecular interactions remain. In the following, we present a systematic investigation of the structure and dynamics of two representative solutions, that is, KCl and LiCl, confined in narrow carbon nanotubes (CNTs) with a diameter of 1.1 and 1.5 nm, using a combination of first-principles and classical MD simulations. Our simulations show that the inclusion of both polarization and cation−π interactions is essential for the description of ion solvation under confinement, particularly for large ions with weak hydration energies. Beyond the variation in ion solvation, we find that cation−π interactions can significantly influence the transport properties of ions in CNTs, particularly for KCl, where our simulations point to a strong correlation between ion dehydration and diffusion. Our study highlights the complex interplay between nanoconfinement and specific intermolecular interactions that strongly control the solvation and transport properties of ions.
Multimodal chromatography is a powerful tool which uses multiple modes of interaction, such as charge and hydrophobicity, to purify protein-based therapeutics. In this work, we performed molecular dynamics simulations of a series of multimodal cation-exchange ligands immobilized on a hydrophilic self-assembled monolayer surface at the commercially relevant surface density (1 ligand/nm2). We found that ligands that were flexible and terminated in a hydrophobic group had a propensity to aggregate on the surface, while less flexible ligands containing a hydrophobic group closer to the surface did not aggregate. For aggregating ligands, this resulted in the formation of a surface pattern that contained relatively large patches of hydrophobicity and charge whose sizes exceeded the length scale of the individual ligands. On the other hand, lowering the surface density to 1 ligand/3 nm2 reduced or eliminated this aggregation behavior. In addition, the introduction of a flexible linker (corresponding to the commercially available ligand) enhanced cluster formation and allowed aggregation to occur at lower surface densities. Further, the use of flexible linkers enabled hydrophobic groups to collapse to the surface, reducing their accessibility. Finally, we developed an approach for quantifying differences in the observed surface patterns by calculating distributions of the patch size and patch length. This clustering phenomenon is likely to play a key role in governing protein–surface interactions in multimodal chromatography. This new understanding of multimodal surfaces has important implications for developing improved predictive models and designing new classes of multimodal separation materials.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.