Journal of Computer-Aided Molecular Design最新文献

African swine fever virus (ASFV), an extremely contagious virus with high mortality rates, causes severe hemorrhagic viral disease in both domestic and wild pigs. Fortunately, ASFV cannot be transmitted from pigs to humans. However, ongoing ASFV outbreaks could have severe economic consequences for global food security. Although ASFV was discovered several years ago, no vaccines or treatments are commercially available yet; therefore, the identification of new anti-ASFV drugs is urgently warranted. Using molecular docking and machine learning, we have previously identified pentagastrin, cangrelor, and fostamatinib as potential antiviral drugs against ASFV. Here, using machine learning combined with docking simulations, we identified natural products with a high affinity for AsfvPolX proteins. We selected five natural products (NPs) that are located close in chemical space to the six known natural flavonoids that possess anti-ASFV activity. Polygalic acid markedly reduced AsfvPolX polymerase activity in a dose-dependent manner. We propose an efficient protocol for identifying NPs as potential antiviral drugs by identifying chemical spaces containing high-affinity binders against ASFV in NP databases.

Drug candidates identified by the pharmaceutical industry typically have unique structural characteristics to ensure they interact strongly and specifically with their biological targets. Identifying these characteristics is a key challenge for developing new drugs, and quantitative structure-activity relationship (QSAR) analysis has generally been used to perform this task. QSAR models with good predictive power improve the cost and time efficiencies invested in compound development. Generating these good models depends on how well differences between “active” and “inactive” compound groups can be conveyed to the model to be learned. Efforts to solve this difference issue have been made, including generating a “molecular descriptor” that compressively expresses the structural characteristics of compounds. From the same perspective, we succeeded in developing the Activity Differences-Quantitative Structure-Activity Relationship (ADis-QSAR) model by generating molecular descriptors that more explicitly convey features of the group through a pair system that performs direct connections between active and inactive groups. We used popular machine learning algorithms, such as Support Vector Machine, Random Forest, XGBoost and Multi-Layer Perceptron for model learning and evaluated the model using scores such as accuracy, area under curve, precision and specificity. The results showed that the Support Vector Machine performed better than the others. Notably, the ADis-QSAR model showed significant improvements in meaningful scores such as precision and specificity compared to the baseline model, even in datasets with dissimilar chemical spaces. This model reduces the risk of selecting false positive compounds, improving the efficiency of drug development.

Kallikrein 6 (KLK6) is an attractive drug target for the treatment of neurological diseases and for various cancers. Herein, we explore the accuracy and efficiency of different computational methods and protocols to predict the free energy of binding (ΔG_bind) for a series of 49 inhibitors of KLK6. We found that the performance of the methods varied strongly with the tested system. For only one of the three KLK6 datasets, the docking scores obtained with rDock were in good agreement (R² ≥ 0.5) with experimental values of ΔG_bind. A similar result was obtained with MM/GBSA (using the ff14SB force field) calculations based on single minimized structures. Improved binding affinity predictions were obtained with the free energy perturbation (FEP) method, with an overall MUE and RMSE of 0.53 and 0.68 kcal/mol, respectively. Furthermore, in a simulation of a real-world drug discovery project, FEP was able to rank the most potent compounds at the top of the list. These results indicate that FEP can be a promising tool for the structure-based optimization of KLK6 inhibitors.

The SAMPL8 blind prediction challenge, which addresses the acid/base dissociation constants (pKa) and the distribution coefficients (logD), was addressed by the Conductor like Screening Model for Realistic Solvation (COSMO-RS). Using the COSMOtherm implementation of COSMO-RS together with a rigorous conformational sampling, yielded logD predictions with a root mean square deviation (RMSD) of 1.36 log units over all 11 compounds and seven bi-phasic systems of the data set, which was the most accurate of all contest submissions (logD).

For the SAMPL8 pKa competition, participants were asked to report the standard state free energies of all microstates, which were then used to calculate the macroscopic pKa. We have used COSMO-RS based linear free energy fit models to calculate the requested energies. The assignment of the calculated and experimental pKa values was made on the basis of the popular transitions, i.e. the transition hat was predicted by the majority of the submissions. With this assignment and a model that covers both, pKa and base pKa, we achieved an RMSD of 3.44 log units (18 pKa values of 14 molecules), which is the second place of the six ranked submissions. By changing to an assignment that is based on the experimental transition curves, the RMSD reduces to 1.65. In addition to the ranked contribution, we submitted two more data sets, one for the standard pKa model and one or the standard base pKa model of COSMOtherm. Using the experiment based assignment with the predictions of the two sets we received a RMSD of 1.42 log units (25 pKa values of 20 molecules). The deviation mainly arises from a single outlier compound, the omission of which leads to an RMSD of 0.89 log units.

As an alternative to antibiotics, Antimicrobial Peptides (AMPs) possess unique properties including cationic, amphipathic and their abundance in nature, but the exact characteristics of AMPs against bacterial membranes are still undetermined. To estimate the structural stability and functional activity of AMPs, the Pseudin AMPs (Pse-1, Pse-2, Pse-3, and Pse-4) from Hylid frog species, Pseudis paradoxa, an abundantly discovered source for AMPs were examined. We studied the intra-peptide interactions and thermal denaturation stability of peptides, as well as the geometrical parameters and secondary structure profiles of their conformational trajectories. On this basis, the peptides were screened out and the highly stable peptide, Pse-4 was subjected to membrane simulation in order to observe the changes in membrane curvature formed by Pse-4 insertion. Monomeric Pse-4 was found to initiate the membrane disruption; however, a stable multimeric form of Pse-4 might be competent to counterbalance the helix-coil transition and to resist the hydrophobic membrane environment. Eventually, hexameric Pse-4 on membrane simulation exhibited the hydrogen bond formation with E. coli bacterial membrane and thereby, leading to the formation of membrane spanning pore that allowed the entry of excess water molecules into the membrane shell, thus causing membrane deformation. Our report points out the mechanism of Pse-4 peptide against the bacterial membrane for the first time. Relatively, Pse-4 works on the barrel stave model against E. coli bacterial membrane; hence it might act as a good therapeutic scaffold in the treatment of multi-drug resistant bacterial strains.

Using generative deep learning models and reinforcement learning together can effectively generate new molecules with desired properties. By employing a multi-objective scoring function, thousands of high-scoring molecules can be generated, making this approach useful for drug discovery and material science. However, the application of these methods can be hindered by computationally expensive or time-consuming scoring procedures, particularly when a large number of function calls are required as feedback in the reinforcement learning optimization. Here, we propose the use of double-loop reinforcement learning with simplified molecular line entry system (SMILES) augmentation to improve the efficiency and speed of the optimization. By adding an inner loop that augments the generated SMILES strings to non-canonical SMILES for use in additional reinforcement learning rounds, we can both reuse the scoring calculations on the molecular level, thereby speeding up the learning process, as well as offer additional protection against mode collapse. We find that employing between 5 and 10 augmentation repetitions is optimal for the scoring functions tested and is further associated with an increased diversity in the generated compounds, improved reproducibility of the sampling runs and the generation of molecules of higher similarity to known ligands.

Computer-aided approaches to ligand design need to balance accuracy with speed. This is particularly true for one of the key parameters to be optimized during ligand development, the free energy of binding ((Delta)G(_{text {bind}})). Here, we developed simple models based on the Linear Interaction Energy approximation to free energy calculation for a G protein-coupled receptor, the serotonin receptor 2A, and critically evaluated their accuracy. Several lessons can be taken from our calculations, providing information on the influence of the docking software used, the conformational state of the receptor, the cocrystallized ligand, and its comparability to the training/test ligands.

Identification of potential therapeutic candidates can be expedited by integrating computational modeling with domain aware machine learning (ML) models followed by experimental validation in an iterative manner. Generative deep learning models can generate thousands of new candidates, however, their physiochemical and biochemical properties are typically not fully optimized. Using our recently developed deep learning models and a scaffold as a starting point, we generated tens of thousands of compounds for SARS-CoV-2 M^pro that preserve the core scaffold. We utilized and implemented several computational tools such as structural alert and toxicity analysis, high throughput virtual screening, ML-based 3D quantitative structure-activity relationships, multi-parameter optimization, and graph neural networks on generated candidates to predict biological activity and binding affinity in advance. As a result of these combined computational endeavors, eight promising candidates were singled out and put through experimental testing using Native Mass Spectrometry and FRET-based functional assays. Two of the tested compounds with quinazoline-2-thiol and acetylpiperidine core moieties showed IC(_{50}) values in the low micromolar range: (2.95pm 0.0017) (upmu)M and 3.41±0.0015 (upmu)M, respectively. Molecular dynamics simulations further highlight that binding of these compounds results in allosteric modulations within the chain B and the interface domains of the M^pro. Our integrated approach provides a platform for data driven lead optimization with rapid characterization and experimental validation in a closed loop that could be applied to other potential protein targets.

An Online tool for Fragment-based Molecule Parametrization (OFraMP) is described. OFraMP is a web application for assigning atomic interaction parameters to large molecules by matching sub-fragments within the target molecule to equivalent sub-fragments within the Automated Topology Builder (ATB, atb.uq.edu.au) database. OFraMP identifies and compares alternative molecular fragments from the ATB database, which contains over 890,000 pre-parameterized molecules, using a novel hierarchical matching procedure. Atoms are considered within the context of an extended local environment (buffer region) with the degree of similarity between an atom in the target molecule and that in the proposed match controlled by varying the size of the buffer region. Adjacent matching atoms are combined into progressively larger matched sub-structures. The user then selects the most appropriate match. OFraMP also allows users to manually alter interaction parameters and automates the submission of missing substructures to the ATB in order to generate parameters for atoms in environments not represented in the existing database. The utility of OFraMP is illustrated using the anti-cancer agent paclitaxel and a dendrimer used in organic semiconductor devices.

Graphical abstract

OFraMP applied to paclitaxel (ATB ID 35922).

Cell division control protein 42 homolog (Cdc42), which controls a variety of cellular functions including rearrangements of the cell cytoskeleton, cell differentiation and proliferation, is a potential cancer therapeutic target. As an endogenous negative regulator of Cdc42, the Rho GDP dissociation inhibitor 1 (RhoGDI1) can prevent the GDP/GTP exchange of Cdc42 to maintain Cdc42 into an inactive state. To investigate the inhibition mechanism of Cdc42 through RhoGDI1 at the atomic level, we performed molecular dynamics (MD) simulations. Without RhoGDI1, Cdc42 has more flexible conformations, especially in switch regions which are vital for binding GDP/GTP and regulators. In the presence of RhoGDI1, it not only can change the intramolecular interactions of Cdc42 but also can maintain the switch regions into a closed conformation through extensive interactions with Cdc42. These results which are consistent with findings of biochemical and mutational studies provide deep structural insights into the inhibition mechanisms of Cdc42 by RhoGDI1. These findings are beneficial for the development of novel therapies targeting Cdc42-related cancers.