This study explores a new approach to pharmacophore screening involving the use of an optimized linear combination of models instead of a single hypothesis. The implementation and evaluation of the developed methodology are performed for a complete known chemical space of 5-HT1AR ligands (3616 active compounds with K i < 100 nM) acquired from the ChEMBL database. Clusters generated from three different methods were the basis for the individual pharmacophore hypotheses, which were assembled into optimal combinations to maximize the different coefficients, namely, MCC, accuracy and recall, to measure the screening performance. Various factors that influence filtering efficiency, including clustering methods, the composition of test sets (random, the most diverse and cluster population-dependent) and hit mode (the compound must fit at least one or two models from a final combination) were investigated. This method outmatched both single hypothesis and random linear combination approaches.

Structural fingerprints and pharmacophore modeling are methodologies that have been used for at least 2 decades in various fields of cheminformatics, from similarity searching to machine learning (ML). Advances in in silico techniques consequently led to combining both these methodologies into a new approach known as the pharmacophore fingerprint. Herein, we propose a high-resolution, pharmacophore fingerprint called Pharmacoprint that encodes the presence, types, and relationships between pharmacophore features of a molecule. Pharmacoprint was evaluated in classification experiments by using ML algorithms (logistic regression, support vector machines, linear support vector machines, and neural networks) and outperformed other popular molecular fingerprints (i.e., ECFP4, Estate, MACCS, PubChem, Substructure, Klekota–Roth, CDK, Extended, and GraphOnly) and the ChemAxon pharmacophoric features fingerprint. Pharmacoprint consisted of 39 973 bits; several methods were applied for dimensionality reduction, and the best algorithm not only reduced the length of the bit string but also improved the efficiency of the ML tests. Further optimization allowed us to define the best parameter settings for using Pharmacoprint in discrimination tests and for maximizing statistical parameters. Finally, Pharmacoprint generated for three-dimensional (3D) structures with defined hydrogens as input data was applied to neural networks with a supervised autoencoder for selecting the most important bits and allowed us to maximize the Matthews correlation coefficient up to 0.962. The results show the potential of Pharmacoprint as a new, perspective tool for computer-aided drug design.

Highly pathogenic influenza viruses pose a serious public health threat to humans. Although vaccines are available, new antivirals are needed to efficiently control disease progression and virus transmission due to the emergence of drug-resistant viral strains. In this study, we describe the anti-viral properties of Soloxolone methyl (SM) (methyl 2-cyano-3,12-dioxo-18βH-olean-9(11),1(2)-dien-30-oate, a chemical derivative of glycyrrhetinic acid) against the flu virus. Anti-flu efficacy studies revealed that SM exhibits antiviral activity against the H1N1 influenza A virus in a dose-dependent manner causing a more than 10-fold decrease in virus titer and a reduction in the expression of NP and M2 viral proteins. In a time-of-addition study, SM was found to act at an early stage of infection to exhibit an inhibitory effect on both the attachment step and virus uptake into cells. Also, in infected cells SM downregulates the expression of the inflammatory cytokines IL-6 and TNF-α. In infected mice, SM administered intranasally prior to and after infection significantly decreases virus titers in the lung and prevents post-challenge pneumonia. Together, these results suggest that Soloxolone methyl might serve as an effective therapeutic agent to manage influenza outbreaks and virus-associated complications, and further preclinical and clinical investigation may be warranted.

Fingerprints, bit representations of compound chemical structure, have been widely used in cheminformatics for many years. Although fingerprints with the highest resolution display satisfactory performance in virtual screening campaigns, the presence of a relatively high number of irrelevant bits introduces noise into data and makes their application more time-consuming. In this study, we present a new method of hybrid reduced fingerprint construction, the Average Information Content Maximization algorithm (AIC-Max algorithm), which selects the most informative bits from a collection of fingerprints. This methodology, applied to the ligands of five cognate serotonin receptors (5-HT2A, 5-HT2B, 5-HT2C, 5-HT5A, 5-HT6), proved that 100 bits selected from four non-hashed fingerprints reflect almost all structural information required for a successful in silico discrimination test. A classification experiment indicated that a reduced representation is able to achieve even slightly better performance than the state-of-the-art 10-times-longer fingerprints and in a significantly shorter time.

No abstract

Despite its remarkable importance in the arena of drug design, serotonin 1A receptor (5-HT1A) has been elusive to the x-ray crystallography community. This lack of direct structural information not only hampers our knowledge regarding the binding modes of many popular ligands (including endogenous neurotransmitter – serotonin), but also limits the search for more potent compounds. In this paper we shed new light on the 3D pharmacological properties of the 5-HT1A receptor by using a ligand-guided approach (ALiBERO) grounded in the Internal Coordinate Mechanics (ICM) docking platform. Starting from a homology template and set of known actives, the method introduces receptor flexibility via Normal Mode Analysis and Monte Carlo sampling, to generate a subset of pockets that display enriched discrimination of actives from inactives in retrospective docking. Here, we thoroughly investigated the repercussions of using different protein templates and the effect of compound selection on screening performance. Finally, the best resulting protein models were applied prospectively in a large virtual screening campaign, in which two new active compounds were identified that were chemically distinct from those described in the literature.

In a search for new anti-HIV-1 chemotypes, we developed a multistep ligand-based virtual screening (VS) protocol combining machine learning (ML) methods with the privileged structures (PS) concept. In its learning step, the VS protocol was based on HIV integrase (IN) inhibitors fetched from the ChEMBL database. The performances of various ML methods and PS weighting scheme were evaluated and applied as VS filtering criteria. Finally, a database of 1.5 million commercially available compounds was virtually screened using a multistep ligand-based cascade, and 13 selected unique structures were tested by measuring the inhibition of HIV replication in infected cells. This approach resulted in the discovery of two novel chemotypes with moderate antiretroviral activity, that, together with their topological diversity, make them good candidates as lead structures for future optimization.

The GABAB receptor (GABAB-R) is a heterodimeric class C G protein-coupled receptor comprised of the GABAB1a/b and GABAB2 subunits. The endogenous orthosteric agonist γ-amino-butyric acid (GABA) binds within the extracellular Venus flytrap (VFT) domain of the GABAB1a/b subunit. The receptor is associated with numerous neurological and neuropsychiatric disorders including learning and memory deficits, depression and anxiety, addiction and epilepsy, and is an interesting target for new drug development. Ligand- and structure-based virtual screening (VS) are used to identify hits in preclinical drug discovery. In the present study, we have evaluated classical ligand-based in silico methods, fingerprinting and pharmacophore mapping and structure-based in silico methods, structure-based pharmacophores, docking and scoring, and linear interaction approximation (LIA) for their aptitude to identify orthosteric GABAB-R compounds. Our results show that the limited number of active compounds and their high structural similarity complicate the use of ligand-based methods. However, by combining ligand-based methods with different structure-based methods active compounds were identified in front of DUDE-E decoys and the number of false positives was reduced, indicating that novel orthosteric GABAB-R compounds may be identified by a combination of ligand-based and structure-based in silico methods.