The third Critical Assessment of Computational Hit-finding Experiments (CACHE) challenged computational teams to identify chemically novel ligands targeting the macrodomain 1 of SARS-CoV-2 Nsp3, a promising coronavirus drug target. Twenty-three groups deployed diverse design strategies to collectively select 1739 ligand candidates. While over 85% of the designed molecules were chemically novel, the best experimentally confirmed hits were structurally similar to previously published compounds. Confirming a trend observed in CACHE #1 and #2, two of the best-performing workflows used compounds selected by physics-based computational screening methods to train machine learning models able to rapidly screen large chemical libraries, while four others used exclusively physics-based approaches. Three pharmacophore searches and one fragment growing strategy were also part of the seven winning workflows. While active molecules discovered by CACHE #3 participants largely mimicked the adenine ring of the endogenous substrate, ADP-ribose, preserving the canonical chemotype commonly observed in previously reported Nsp3-Mac1 ligands, they still provide novel structure-activity relationship insights that may inform the development of future antivirals. Collectively, these results show that multiple molecular design strategies can efficiently converge on similar potent molecules.

A critical assessment of computational hit-finding experiments (CACHE) challenge was conducted to predict ligands for the SARS-CoV-2 Nsp13 helicase RNA binding site, a highly conserved COVID-19 target. Twenty-three participating teams comprised of computational chemists and data scientists used protein structure and data from fragment-screening paired with advanced computational and machine learning methods to each predict up to 100 inhibitory ligands. Across all teams, 1957 compounds were predicted and were subsequently procured from commercial catalogs for biophysical assays. Of these compounds, 0.7% were confirmed to bind to Nsp13 in a surface plasmon resonance assay. The six best-performing computational workflows used fragment growing, active learning, or conventional virtual screening with and without complementary deep-learning scoring functions. Follow-up functional assays resulted in identification of two compound scaffolds that bound Nsp13 with a Kd below 10 μM and inhibited in vitro helicase activity. Overall, CACHE #2 participants were successful in identifying hit compound scaffolds targeting Nsp13, a central component of the coronavirus replication-transcription complex. Computational design strategies recurrently successful across the first two CACHE challenges include linking or growing docked or crystallized fragments and docking small and diverse libraries to train ultrafast machine-learning models. The CACHE #2 competition reveals how crowd-sourcing ligand prediction efforts using a distinct array of approaches followed with critical biophysical assays can result in novel lead compounds to advance drug discovery efforts.

Active learning (AL) can significantly accelerate drug discovery by iteratively selecting informative molecules, reducing experimental workload. However, existing AL studies typically assume access to large datasets, an unrealistic scenario for most academic labs. AL strategies tailored specifically for small-scale molecular screening, are investigated, using only 110 affinity evaluations approximated by docking scores from realistic compound libraries: the Developmental Therapeutics Program repository (DTP) and Enamine Discovery Diversity Set 10 (DDS-10). Among 20 tested combinations of molecular descriptors and machine learning models, continuous and data-driven descriptors combined with a multilayer perceptron, augmented by the pairwise difference regression data augmentation technique, are identified as optimal. This combination achieves a 100% probability of discovering at least five top-1% hits from DTP using only 110 affinity evaluations which remains high under simulated experimental uncertainty. Similarly, the DDS-10 dataset achieves a 100% probability of discovering at least five top-1% hits. Incorporating prior knowledge by enriching initial datasets with a single known hit molecule increased the probability of finding 20 or more hits. These findings underscore the feasibility and substantial potential of AL for small-scale drug discovery in resource-limited environments. These results suggest that early in the AL search the algorithm benefits from accurately quantifying the binding strengths of very weak binders.

Pd EnCat™ 30 is a palladium catalyst broadly used in several hydrogenation and cross-coupling reactions. It is known for its numerous beneficial features, which include high-yielding performance, easy recovery, and reusability. However, the available data regarding its recyclability in Suzuki coupling reactions are limited to a few reaction cycles and, therefore, fail to explore its full potential. Our work focuses on investigating the extent of Pd EnCat™ 30 reusability in Suzuki cross-coupling reactions by measuring its performance according to isolated yields of product. Our findings demonstrate that Pd EnCat™ 30 can be reused over a minimum of 30 reaction cycles, which is advantageous in terms of cost reduction and more sustainable chemical production.

Steroid hormones play a crucial role in the body by acting as chemical messengers. They are, however, poorly soluble in water, and cyclodextrins can increase their solubility thus leading to increased bioavailability when used in drug formulations. Accuracy in the prediction of the free energy of binding of cyclodextrin/steroid inclusion complexes with simulation is important because of the potential value it brings by providing low-cost predictions of the real-life behavior of the cyclodextrin/steroid inclusion complex and the potential for high-through-put screening. Many computational methods exist, and it is therefore important to understand the ability of current theoretical models to accurately predict the free energy of binding for these inclusion complexes. We focused specifically on the estimation of the free energy of binding of inclusion complexes of four steroids: Hydrocortisone, dexamethasone, prednisolone, and 6α-methylprednisolone with native α-CD, β-CD, γ-CD, (2-hydroxy)propyl-β-CD, and sulfobutylether-β-CD by phase solubility as well as with α, β, and γ-CD by simulations. The simulations were assessed with both docking and the molecular mechanics combined with the generalized Born and surface area (MM/GBSA) continuum solvation approach. Considering the phase solubility diagram, (2-hydroxy)propyl-β-CD and sulfobutylether-β-CD dissolved more steroids in the higher concentration range as expected. The assessment of the free energy of binding obtained from the phase solubility and theory showed that the MM/GBSA method has shown promise in reliably generating accurate predictions in the field of calculating the free energy of binding of steroids/cyclodextrins with a correlation coefficient (R2) = 0.94.

The CACHE challenges are a series of prospective benchmarking exercises to evaluate progress in the field of computational hit-finding. Here we report the results of the inaugural CACHE challenge in which 23 computational teams each selected up to 100 commercially available compounds that they predicted would bind to the WDR domain of the Parkinson's disease target LRRK2, a domain with no known ligand and only an apo structure in the PDB. The lack of known binding data and presumably low druggability of the target is a challenge to computational hit finding methods. Of the 1955 molecules predicted by participants in Round 1 of the challenge, 73 were found to bind to LRRK2 in an SPR assay with a KD lower than 150 μM. These 73 molecules were advanced to the Round 2 hit expansion phase, where computational teams each selected up to 50 analogs. Binding was observed in two orthogonal assays for seven chemically diverse series, with affinities ranging from 18 to 140 μM. The seven successful computational workflows varied in their screening strategies and techniques. Three used molecular dynamics to produce a conformational ensemble of the targeted site, three included a fragment docking step, three implemented a generative design strategy and five used one or more deep learning steps. CACHE #1 reflects a highly exploratory phase in computational drug design where participants adopted strikingly diverging screening strategies. Machine learning-accelerated methods achieved similar results to brute force (e.g., exhaustive) docking. First-in-class, experimentally confirmed compounds were rare and weakly potent, indicating that recent advances are not sufficient to effectively address challenging targets.

Interoperability in computational chemistry is elusive, impeded by the independent development of software packages and idiosyncratic nature of their output files. The cclib library was introduced in 2006 as an attempt to improve this situation by providing a consistent interface to the results of various quantum chemistry programs. The shared API across programs enabled by cclib has allowed users to focus on results as opposed to output and to combine data from multiple programs or develop generic downstream tools. Initial development, however, did not anticipate the rapid progress of computational capabilities, novel methods, and new programs; nor did it foresee the growing need for customizability. Here, we recount this history and present cclib 2, focused on extensibility and modularity. We also introduce recent design pivots—the formalization of cclib’s intermediate data representation as a tree-based structure, a new combinator-based parser organization, and parsed chemical properties as extensible objects.

Drug-resistant bacterial wound infections have become a major threat to human health worldwide, and there is an urgent need to develop a new generation of antibacterial agents to replace conventional antibiotics. In this work, we proposed an efficient nanoplatform that combines photothermal therapy (PTT) and light-triggered release of nitric oxide (NO) to combat bacteria. A multifunctional nanoplatform (BBDH NPs) based on a BODIPY probe, NO thermal responsive donor [N, N′-di-sec-butyl-N, N′-diniroso-1,4-phenylenediamine (BNN6)], and a PEGylated polymer was prepared with a nanoprecipitation method. BBDH NPs represent a one-two punch against bacterial infections, combining potent photothermal therapy and the controlled release of NO, enabling rapid and efficient eradication of gram-negative and gram-positive bacteria. Histological analysis on a mouse model demonstrates that wounds treated with BBDH NPs and 685 nm laser irradiation have completed re-epithelialization, significant collagen deposition, and a number of hair follicle formation. BBDH NPs also exhibit a remarkable therapeutic effect on wounds infected with methicillin-resistant Staphylococcus aureus (MRSA). These results highlight that the proposed synergistic antibacterial strategy can be used as a potential therapeutic tool in drug-resistant, bacterial-infected wounds.

In immunoglobulin light chain amyloidosis, unstable immunoglobulin light chains aggregate as amyloid fibrils in organs, leading to organ failure and death in more than 50% of untreated patients. Research by Morgan et al. (Morgan, G. J. et. al., Proceedings of the National Academy of Sciences 2019, 116, 8360–8369) and Yan et al. (Yan, N. L. et. al., Journal of Medicinal Chemistry 2021, 64, 6273–6299) has shown that several classes of small molecules can be used as kinetic stabilizers of native light chains and potentially slow or stop the disease. Based on their criteria of a clinically successful kinetic stabilizer, this study includes an in-silico virtual screening of drugs approved by the U.S. Food and Drug Administration to find promising candidates of kinetic stabilizers through drug repurposing. Of these candidates, molecular dynamics simulations and subsequent studies of molecular mechanics with generalized Born and surface area solvation suggest that delavirdine and pazopanib can potentially be used as kinetic stabilizers of native light chains.

Self-driving labs (SDLs) leverage combinations of artificial intelligence, automation, and advanced computing to accelerate scientific discovery.

Veratridine (VTD) is a plant neurotoxin that acts by blocking the voltage-gated sodium channels (VGSC) of cell membranes. Symptoms of VTD intoxication include intense nausea, hypotension, arrhythmia, and loss of consciousness. The treatment for the intoxication is mainly focused on treating the symptoms, meaning there is no specific antidote against VTD. In this pursuit, we were interested in studying the molecular interactions of VTD with cyclodextrins (CDs). CDs are supramolecular macrocycles with the ability to form host–guest inclusion complexes (ICs) inside their hydrophobic cavity. Since VTD is a lipid-soluble alkaloid, we hypothesized that it could form stable inclusion complexes with different types of CDs, resulting in changes to its physicochemical properties. In this investigation, we studied the interaction of VTD with β-CD, γ-CD and sulfobutyl ether β-CD (SBCD) by isothermal titration calorimetry (ITC) and nuclear magnetic resonance (NMR) spectroscopy. Docking and molecular dynamics studies confirmed the most stable configuration for the inclusion complexes. Finally, with an interest in understanding the effects of the VTD/CD molecular interactions, we performed cell-based assays (CBAs) on Neuro-2a cells. Our findings reveal that the use of different amounts of CDs has an antidote-like concentration-dependent effect on the cells, significantly increasing cell viability and thus opening opportunities for novel research on applications of CDs and VTD.

2-Hydroxypropyl-β-cyclodextrins (HP-β-CDs) ability to form inclusion complexes with lipophilic compounds is important both to treat Niemann Pick Type C disease and as part of a more general drug-formulation as a drug carrier. Theoretical studies of inclusion complexes with classical molecular dynamics simulations of HP-β-CDs are hampered by the fact that there are more than 2 million possible isomers to consider. We construct a general GLYCAM06 (Kirschner, K. N. et. al., J. Comput. Chem. 2007, 29, 622–655) compatible force field to treat all possible isomers of HP-β-CDs and investigate the effect on free energies of binding obtained from the MM/GBSA approach when increasing the degree of substitution. We find improved binding proportional to the added side chains. To show the applicability of the HP-β-CD force field, we generate an ensemble of simulations guided by experimentally determined distributions of isomers from mass spectrometry. Our results for ibuprofen and ketoprofen shows that acceptable accuracy in free energies of binding can be obtained with ensembles generated from around 40 isomers, but this is dependent on the guest molecule. We also investigate a simple linear relation between molecular log(P ) and free energy of binding for inclusion complexes of HP-β-CD. The linear model is promising but we find that the accuracy is too dependent on a single simulation. The presented force field and computational approach should provide for more realistic simulations.

The aug-cc-pVTZ-J basis set family is extended to include the fourth-row p-block elements Ga, Ge, As, Se, and Br. We use the established approach outlined by Sauer and coworkers (J. Chem. Phys. 115, 1324 [2001], J. Chem. Phys. 133, 054308 [2010], J. Chem. Theory Comput. 7, 4070 [2011], and J. Chem. Theory Comput. 7, 4077 [2011]) where the completely uncontracted aug-cc-pVTZ basis set is saturated with tight s-, p-, d-, and f-functions to form the aug-cc-pVTZ-Juc basis set for the tested elements. The saturation is carried out on the simplest hydrides possible for the tested elements GaH, GeH4, AsH3, H2Se, and HBr until an improvement is less than 0.01% for all s-, p-, and d-functions added. f-Functions are added to an improvement less than or equal to 1.0% due to the computational expense these functions add. The saturated aug-cc-pVTZ-Juc (26s16p12d5f) is then recontracted using the molecular orbital coefficients from self-consistent field calculations on the simple hydrides to improve computational efficiency. During contraction of the basis set, we observe that the linear hydrogen bromide molecule has a slower convergence than the other tested molecules which sets a limit on the accuracy obtained. All calculations with the contracted aug-cc-pVTZ-J [17s10p7d5f] gives results that are within 1.0% of the uncontracted results at considerable computational savings.

A graph-based genetic algorithm (GA) is used to identify molecules (ligands) with high absolute docking scores as estimated by the Glide software, starting from randomly chosen molecules from the ZINC database, for four different targets: Bacillus subtilis chorismate mutase (CM), human β2-adrenergic G protein-coupled receptor (β2AR), the DDR1 kinase domain (DDR1), and β-cyclodextrin (BCD). By the combined use of functional group filters and a score modifier based on a heuristic synthetic accessibility (SA) score our approach identifies between ca 500 and 6000 structurally diverse molecules with scores better than known binders by screening a total of 400,000 molecules starting from 8000 randomly selected molecules from the ZINC database. Screening 250,000 molecules from the ZINC database identifies significantly more molecules with better docking scores than known binders, with the exception of CM, where the conventional screening approach only identifies 60 compounds compared to 511 with GA+Filter+SA. In the case of β2AR and DDR1 the GA+Filter+SA approach finds significantly more molecules with docking scores lower than -9.0 and -10.0. The GA+Filters+SA docking methodology is thus effective in generating a large and diverse set of synthetically accessible molecules with very good docking scores for a particular target. An early incarnation of the GA+Filter+SA approach was used to identify potential binders to the COVID-19 main protease and submitted to the early stages of the COVID Moonshot project, a crowd-sourced initiative to accelerate the development of a COVID antiviral.

The fragment-based polarizable embedding model combined with an appropriate electronic-structure method constitutes a highly efficient and accurate multiscale approach for computing spectroscopic properties of a central moiety including effects from its molecular environment through an embedding potential. There is, however, a comparatively high computational overhead associated with the computation of the embedding potential which is derived from first principles calculations on individual fragments of the environment. To reduce the computational cost associated with the calculation of embedding-potential parameters, we developed a set of amino-acid-specific transferable parameters tailored for large-scale polarizable embedding calculations that include proteins. The amino-acid-based parameters are obtained by simultaneously fitting to a set of reference electric potentials based on structures derived from a backbone-dependent rotamer library. The developed cost-effective polarizable protein potential (CP3) consists of atom-centered charges and isotropic dipole-dipole polarizabilities of the standard amino acids. In terms of reproduction of electric potentials, the CP3 is shown to perform consistently and with acceptable accuracy across both small tripeptide test systems and larger proteins. We show, through applications on realistic protein systems, that acceptable accuracy can be obtained by using a pure CP3 representation of the protein environment, thus altogether omitting the cost associated with the calculation of embedding-potential parameters. High accuracy comparable to the full fragment-based approach can be achieved through a mixed description where the CP3 is used only to describe amino acids beyond a threshold distance from the central quantum part.

The polarizable embedding (PE) model is a fragment-based quantum-classical approach aimed at accurate inclusion of environment effects in quantum-mechanical response property calculations. The aim of this tutorial is to give insight into the practical use of the PE model. Starting from a set of molecular structures and until you arrive at the final property, there are many crucial details to consider in order to obtain trustworthy results in an efficient manner. To lower the threshold for new users wanting to explore the use of the PE model, we describe and discuss important aspects related to its practical use. This includes directions on how to generate input files and how to run a calculation.

Modeling of spectral properties of extended chemical systems, such as the case of a solute in a solvent, is often performed based on so-called hybrid models in which only part of the complete system is given a quantum chemical description. The remaining part of the system is represented by an embedding potential treating the environment either by a discrete or continuum model. In order to successfully make use of minimally sized quantum chemical regions, the embedding potential should represent the environment as authentic as possible. Here, the importance of exactly such an accurate description of the embedding potential is investigated by comparing the performance of the polarizable embedding scheme against larger-sized full quantum mechanical calculations. Our main conclusion is that as long as the solute and solvent do not overlap in their absorption spectra, the polarizable embedding approach shows results consistent with full quantum chemical calculations. For partly overlapping absorption spectra, the polarizable embedding approach can furthermore successfully be expanded within a Frenkel exciton approach based on only economical monomeric quantum chemical calculations. Thus, by extending the polarizable embedding scheme to the exciton picture, it is possible to cover computations of the whole absorption spectrum and still reduce the computational cost compared to costly cluster calculations.

We describe the a new molecular simulation package that is designed for ab initio molecular dynamics simulations of molecular and condensed-phase chemical reactions and otherprocesses, with particular focus on mean-field and quantum embedding methods for electronic structure.