Molecular fingerprint similarity search in virtual screening

Molecular fingerprints have been used for a long time now in drug discovery and virtual screening. Their ease of use (requiring little to no configuration) and the speed at which substructure and similarity searches can be performed with them - paired with a virtual screening performance similar to other more complex methods - is the reason for their popularity. However, there are many types of fingerprints, each representing a different aspect of the molecule, which can greatly affect search performance. This review focuses on commonly used fingerprint algorithms, their usage in virtual screening, and the software packages and online tools that provide these algorithms.

The Light and Dark Sides of Virtual Screening: What Is There to Know?

Virtual screening consists of using computational tools to predict potentially bioactive compounds from files containing large libraries of small molecules. Virtual screening is becoming increasingly popular in the field of drug discovery as in silico techniques are continuously being developed, improved, and made available. As most of these techniques are easy to use, both private and public organizations apply virtual screening methodologies to save resources in the laboratory. However, it is often the case that the techniques implemented in virtual screening workflows are restricted to those that the research team knows. Moreover, although the software is often easy to use, each methodology has a series of drawbacks that should be avoided so that false results or artifacts are not produced. Here, we review the most common methodologies used in virtual screening workflows in order to both introduce the inexperienced researcher to new methodologies and advise the experienced researcher on how to prevent common mistakes and the improper usage of virtual screening methodologies.

DecoyFinder: an easy-to-use python GUI application for building target-specific decoy sets

Decoys are molecules that are presumed to be inactive against a target (i.e. will not likely bind to the target) and are used to validate the performance of molecular docking or a virtual screening workflow. The Directory of Useful Decoys database (http://dud.docking.org/) provides a free directory of decoys for use in virtual screening, though it only contains a limited set of decoys for 40 targets.To overcome this limitation, we have developed an application called DecoyFinder that selects, for a given collection of active ligands of a target, a set of decoys from a database of compounds. Decoys are selected if they are similar to active ligands according to five physical descriptors (molecular weight, number of rotational bonds, total hydrogen bond donors, total hydrogen bond acceptors and the octanol-water partition coefficient) without being chemically similar to any of the active ligands used as an input (according to the Tanimoto coefficient between MACCS fingerprints). To the best of our knowledge, DecoyFinder is the first application designed to build target-specific decoy sets.

AVAILABILITY

A complete description of the software is included on the application home page. A validation of DecoyFinder on 10 DUD targets is provided as Supplementary Table S1. DecoyFinder is freely available at http://URVnutrigenomica-CTNS.github.com/DecoyFinder.

Prediction of Novel Inhibitors of the Main Protease (M-pro) of SARS-CoV-2 through Consensus Docking and Drug Reposition

Since the outbreak of the COVID-19 pandemic in December 2019 and its rapid spread worldwide, the scientific community has been under pressure to react and make progress in the development of an effective treatment against the virus responsible for the disease. Here, we implement an original virtual screening (VS) protocol for repositioning approved drugs in order to predict which of them could inhibit the main protease of the virus (M-pro), a key target for antiviral drugs given its essential role in the virus' replication. Two different libraries of approved drugs were docked against the structure of M-pro using Glide, FRED and AutoDock Vina, and only the equivalent high affinity binding modes predicted simultaneously by the three docking programs were considered to correspond to bioactive poses. In this way, we took advantage of the three sampling algorithms to generate hypothetic binding modes without relying on a single scoring function to rank the results. Seven possible SARS-CoV-2 M-pro inhibitors were predicted using this approach: Perampanel, Carprofen, Celecoxib, Alprazolam, Trovafloxacin, Sarafloxacin and ethyl biscoumacetate. Carprofen and Celecoxib have been selected by the COVID Moonshot initiative for in vitro testing; they show 3.97 and 11.90% M-pro inhibition at 50 µM, respectively.

Haste makes waste: A critical review of docking-based virtual screening in drug repurposing for SARS-CoV-2 main protease (M-pro) inhibition

This review makes a critical evaluation of 61 peer‐reviewed manuscripts that use a docking step in a virtual screening (VS) protocol to predict SARS‐CoV‐2 M‐pro (M‐pro) inhibitors in approved or investigational drugs. Various manuscripts predict different compounds, even when they use a similar initial dataset and methodology, and most of them do not validate their methodology or results. In addition, a set of known 150 SARS‐CoV‐2 M‐pro inhibitors extracted from the literature and a second set of 81 M‐pro inhibitors and 113 inactive compounds obtained from the COVID Moonshot project were used to evaluate the reliability of using docking scores as feasible predictors of the potency of a SARS‐CoV‐2 M‐pro inhibitor. Using two SARS‐CoV‐2 M‐pro structures and five protein‐ligand docking programs, we proved that the correlation between the pIC₅₀ and docking scores is not good. Neither was any correlation found between the pIC₅₀ and the ∆G calculated with an MM‐GBSA method. When a group of experimentally known inactive compounds was added, neither the docking scores or the ∆G were able to distinguish between compounds with or without M‐pro experimental inhibitory activity. Performances improved when covalent and noncovalent inhibitors were treated separately, but were not good enough to fully support using a docking score as a cutoff value for selecting new putative M‐pro inhibitors or predicting the relative bioactivity of a compound by comparison with a reference compound. The two sets of known SARS‐CoV‐2 M‐pro inhibitors presented here could be used for validating future VS protocols which aim to predict M‐pro inhibitors.

The good, the bad and the dubious: VHELIBS, a validation helper for ligands and binding sites

BACKGROUND

Many Protein Data Bank (PDB) users assume that the deposited structural models are of high quality but forget that these models are derived from the interpretation of experimental data. The accuracy of atom coordinates is not homogeneous between models or throughout the same model. To avoid basing a research project on a flawed model, we present a tool for assessing the quality of ligands and binding sites in crystallographic models from the PDB.

RESULTS

The Validation HElper for LIgands and Binding Sites (VHELIBS) is software that aims to ease the validation of binding site and ligand coordinates for non-crystallographers (i.e., users with little or no crystallography knowledge). Using a convenient graphical user interface, it allows one to check how ligand and binding site coordinates fit to the electron density map. VHELIBS can use models from either the PDB or the PDB_REDO databank of re-refined and re-built crystallographic models. The user can specify threshold values for a series of properties related to the fit of coordinates to electron density (Real Space R, Real Space Correlation Coefficient and average occupancy are used by default). VHELIBS will automatically classify residues and ligands as Good, Dubious or Bad based on the specified limits. The user is also able to visually check the quality of the fit of residues and ligands to the electron density map and reclassify them if needed.

CONCLUSIONS

VHELIBS allows inexperienced users to examine the binding site and the ligand coordinates in relation to the experimental data. This is an important step to evaluate models for their fitness for drug discovery purposes such as structure-based pharmacophore development and protein-ligand docking experiments.

Activity and selectivity cliffs for DPP-IV inhibitors: Lessons we can learn from SAR studies and their application to virtual screening

The inhibition of dipeptidyl peptidase-IV (DPP-IV) has emerged over the last decade as one of the most effective treatments for type 2 diabetes mellitus, and consequently (a) 11 DPP-IV inhibitors have been on the market since 2006 (three in 2015), and (b) 74 noncovalent complexes involving human DPP-IV and drug-like inhibitors are available at the Protein Data Bank (PDB). The present review aims to (a) explain the most important activity cliffs for DPP-IV noncovalent inhibition according to the binding site structure of DPP-IV, (b) explain the most important selectivity cliffs for DPP-IV noncovalent inhibition in comparison with other related enzymes (i.e., DPP8 and DPP9), and (c) use the information deriving from this activity/selectivity cliff analysis to suggest how virtual screening protocols might be improved to favor the early identification of potent and selective DPP-IV inhibitors in molecular databases (because they have not succeeded in identifying selective DPP-IV inhibitors with IC₅₀ ≤ 100 nM). All these goals are achieved with the help of available homology models for DPP8 and DPP9 and an analysis of the structure-activity studies used to develop the noncovalent inhibitors that form part of some of the complexes with human DPP-IV available at the PDB.

Identification of novel human dipeptidyl peptidase-IV inhibitors of natural origin (part I): virtual screening and activity assays

Background

There has been great interest in determining whether natural products show biological activity toward protein targets of pharmacological relevance. One target of particular interest is DPP-IV whose most important substrates are incretins that, among other beneficial effects, stimulates insulin biosynthesis and secretion. Incretins have very short half-lives because of their rapid degradation by DPP-IV and, therefore, inhibiting this enzyme improves glucose homeostasis. As a result, DPP-IV inhibitors are of considerable interest to the pharmaceutical industry. The main goals of this study were (a) to develop a virtual screening process to identify potential DPP-IV inhibitors of natural origin; (b) to evaluate the reliability of our virtual-screening protocol by experimentally testing the in vitro activity of selected natural-product hits; and (c) to use the most active hit for predicting derivatives with higher binding affinities for the DPP-IV binding site.

Methodology/Principal Findings

We predicted that 446 out of the 89,165 molecules present in the natural products subset of the ZINC database would inhibit DPP-IV with good ADMET properties. Notably, when these 446 molecules were merged with 2,342 known DPP-IV inhibitors and the resulting set was classified into 50 clusters according to chemical similarity, there were 12 clusters that contained only natural products for which no DPP-IV inhibitory activity has been previously reported. Nine molecules from 7 of these 12 clusters were then selected for in vitro activity testing and 7 out of the 9 molecules were shown to inhibit DPP-IV (where the remaining two molecules could not be solubilized, preventing the evaluation of their DPP-IV inhibitory activity). Then, the hit with the highest activity was used as a lead compound in the prediction of more potent derivatives.

Conclusions/Significance

We have demonstrated that our virtual-screening protocol was successful in identifying novel lead compounds for developing more potent DPP-IV inhibitors.

The Mutational Landscape of SARS-CoV-2

Mutation research is crucial for detecting and treating SARS-CoV-2 and developing vaccines. Using over 5,300,000 sequences from SARS-CoV-2 genomes and custom Python programs, we analyzed the mutational landscape of SARS-CoV-2. Although almost every nucleotide in the SARS-CoV-2 genome has mutated at some time, the substantial differences in the frequency and regularity of mutations warrant further examination. C>U mutations are the most common. They are found in the largest number of variants, pangolin lineages, and countries, which indicates that they are a driving force behind the evolution of SARS-CoV-2. Not all SARS-CoV-2 genes have mutated in the same way. Fewer non-synonymous single nucleotide variations are found in genes that encode proteins with a critical role in virus replication than in genes with ancillary roles. Some genes, such as spike (S) and nucleocapsid (N), show more non-synonymous mutations than others. Although the prevalence of mutations in the target regions of COVID-19 diagnostic RT-qPCR tests is generally low, in some cases, such as for some primers that bind to the N gene, it is significant. Therefore, ongoing monitoring of SARS-CoV-2 mutations is crucial. The SARS-CoV-2 Mutation Portal provides access to a database of SARS-CoV-2 mutations.

Anti-inflammatory and Proapoptotic Properties of the Natural Compound o-Orsellinaldehyde

Metabolic syndrome is a cluster of medical conditions that increases the risk of developing cardiovascular disease and type 2 diabetes. Numerous studies have shown that inflammation is directly involved in the onset of metabolic syndrome and related pathologies. In this study, in silico techniques were applied to a natural products database containing molecules isolated from mushrooms from the Catalan forests to predict molecules that can act as human nuclear-factor κβ kinase 2 (IKK-2) inhibitors. IKK-2 is the main component responsible for activating the nuclear-factor κβ transcription factor (NF-κβ). One of these predicted molecules was o-orsellinaldehyde, a molecule present in the mushroom Grifola frondosa. This study shows that o-orsellinaldehyde presents anti-inflammatory and pro-apoptotic properties by acting as IKK-2 inhibitor. Additionally, we suggest that the anti-inflammatory and pro-apoptotic properties of Grifola frondosa mushroom could partially be explained by the presence of o-orsellinaldehyde on its composition.

Gut Microbiome and Small RNA Integrative-Omic Perspective of Meconium and Milk-FED Infant Stool Samples

The human gut microbiome plays an important role in health, and its initial development is conditioned by many factors, such as feeding. It has also been claimed that this colonization is guided by bacterial populations, the dynamic virome, and transkingdom interactions between host and microbial cells, partially mediated by epigenetic signaling. In this article, we characterized the bacteriome, virome, and smallRNome and their interaction in the meconium and stool samples from infants. Bacterial and viral DNA and RNA were extracted from the meconium and stool samples of 2- to 4-month-old milk-fed infants. The bacteriome, DNA and RNA virome, and smallRNome were assessed using 16S rRNA V4 sequencing, viral enrichment sequencing, and small RNA sequencing protocols, respectively. Data pathway analysis and integration were performed using the R package mixOmics. Our findings showed that the bacteriome differed among the three groups, while the virome and smallRNome presented significant differences, mainly between the meconium and stool of milk-fed infants. The gut environment is rapidly acquired after birth, and it is highly adaptable due to the interaction of environmental factors. Additionally, transkingdom interactions between viruses and bacteria can influence host and smallRNome profiles. However, virome characterization has several protocol limitations that must be considered.

Prediction of Recurrent Mutations in SARS-CoV-2 Using Artificial Neural Networks

Predicting SARS-CoV-2 mutations is difficult, but predicting recurrent mutations driven by the host, such as those caused by host deaminases, is feasible. We used machine learning to predict which positions from the SARS-CoV-2 genome will hold a recurrent mutation and which mutations will be the most recurrent. We used data from April 2021 that we separated into three sets: a training set, a validation set, and an independent test set. For the test set, we obtained a specificity value of 0.69, a sensitivity value of 0.79, and an Area Under the Curve (AUC) of 0.8, showing that the prediction of recurrent SARS-CoV-2 mutations is feasible. Subsequently, we compared our predictions with updated data from January 2022, showing that some of the false positives in our prediction model become true positives later on. The most important variables detected by the model's Shapley Additive exPlanation (SHAP) are the nucleotide that mutates and RNA reactivity. This is consistent with the SARS-CoV-2 mutational bias pattern and the preference of some host deaminases for specific sequences and RNA secondary structures. We extend our investigation by analyzing the mutations from the variants of concern Alpha, Beta, Delta, Gamma, and Omicron. Finally, we analyzed amino acid changes by looking at the predicted recurrent mutations in the M-pro and spike proteins.