Analysing and Navigating Natural Products Space for Generating Small, Diverse, But Representative Chemical Libraries

De novo generation of dual-target ligands for the treatment of SARS-CoV-2 using deep learning, virtual screening, and molecular dynamic simulations

De novo generation of molecules with the necessary features offers a promising opportunity for artificial intelligence, such as deep generative approaches. However, creating novel compounds having biological activities toward two distinct targets continues to be a very challenging task. In this study, we develop a unique computational framework for the de novo synthesis of bioactive compounds directed at two predetermined therapeutic targets. This framework is referred to as the dual-target ligand generative network. Our approach uses a stochastic policy to explore chemical spaces called a sequence-based simple molecular input line entry system (SMILES) generator. The steps in the high-level workflow would be to gather and prepare the training data for both targets' molecules, build a neural network model and train it to make molecules, create new molecules using generative AI, and then virtually screen the newly validated molecules against the SARS-CoV-2 PLpro and 3CLpro drug targets. Results shows that novel molecules generated have higher binding affinity with both targets than the conventional drug i.e. Remdesivir being used for the treatment of SARS-CoV-2.Communicated by Ramaswamy H. Sarma.

Navigating the Expansive Landscapes of Soft Materials: A User Guide for High-Throughput Workflows

Synthetic polymers are highly customizable with tailored structures and functionality, yet this versatility generates challenges in the design of advanced materials due to the size and complexity of the design space. Thus, exploration and optimization of polymer properties using combinatorial libraries has become increasingly common, which requires careful selection of synthetic strategies, characterization techniques, and rapid processing workflows to obtain fundamental principles from these large data sets. Herein, we provide guidelines for strategic design of macromolecule libraries and workflows to efficiently navigate these high-dimensional design spaces. We describe synthetic methods for multiple library sizes and structures as well as characterization methods to rapidly generate data sets, including tools that can be adapted from biological workflows. We further highlight relevant insights from statistics and machine learning to aid in data featurization, representation, and analysis. This Perspective acts as a "user guide" for researchers interested in leveraging high-throughput screening toward the design of multifunctional polymers and predictive modeling of structure-property relationships in soft materials.

In silico structural, phylogenetic and drug target analysis of putrescine monooxygenase from Shewanella putrefaciens-95

BACKGROUND

The enormous and irresponsible use of antibiotics has led to the emergence of resistant strains of bacteria globally. A new approach to combat this crisis has been nutritional immunity limiting the availability of nutrients to pathogens. Targeting the siderophore biosynthetic pathway that helps in iron acquisition, an essential microelement in the bacterial system has been the topic of interest in recent days that backs the concept of nutritional immunity. Supporting this view, we have chosen to study a key enzyme in the biosynthetic pathway of putrebactin called putrescine monooxygenase (SpPMO) from Shewanella putrefaciens. In our previous study, we co-expressed putrescine monooxygenase recombinantly in Escherichia coli BL21 Star (DE3). The bioinformatic analysis and screening of inhibitors will broaden the scope of SpPMO as a drug target.

RESULTS

In the present study, we have analysed the physicochemical properties of the target enzyme and other N-hydroxylating monooxygenases (NMOs) using ExPASy server. The target enzyme SpPMO and most of the selected NMOs have a slightly acidic isoelectric point and are medially thermostable and generally insoluble. The multiple sequence alignment identified the GXGXX(N/A), DXXXFATGYXXXXP motives and conserved amino acids involved in FAD binding, NADP binding, secondary structure formation and substrate binding. The phylogenetic analysis indicated the distribution of the monooxygenases into different clades according to their substrate specificity. Further, a 3D model of SpPMO was predicted using I-TASSER online tool with DfoA from Erwinia amylovora as a template. The model was validated using the SAVES server and deposited to the Protein Model Database with the accession number PM0082222. The molecular docking analysis with different substrates revealed the presence of a putrescine binding pocket made of conserved amino acids and another binding pocket present on the surface of the protein wherein all other ligands interact with high binding affinity. The molecular docking of naturally occurring inhibitor molecules with SpPMO 3D model identified curcumin and niazirin with 1.83 and 2.81 μM inhibition constants as two promising inhibitors. Further studies on kinetic parameters of curcumin and niazirin inhibitors in vitro determined the Ki to be 2.6±0.0036 μM and 18.38±0.008 μM respectively.

CONCLUSION

This analysis will help us understand the structural, phylogenetic and drug target aspects of putrescine monooxygenase from Shewanella putrefaciens-95 in detail. It sheds light on the precautionary measures that can be developed to inhibit the enzyme and thereby the secondary infections caused by them.

De novo generation of dual-target ligands using adversarial training and reinforcement learning

Artificial intelligence, such as deep generative methods, represents a promising solution to de novo design of molecules with the desired properties. However, generating new molecules with biological activities toward two specific targets remains an extremely difficult challenge. In this work, we conceive a novel computational framework, herein called dual-target ligand generative network (DLGN), for the de novo generation of bioactive molecules toward two given objectives. Via adversarial training and reinforcement learning, DLGN treats a sequence-based simplified molecular input line entry system (SMILES) generator as a stochastic policy for exploring chemical spaces. Two discriminators are then used to encourage the generation of molecules that belong to the intersection of two bioactive-compound distributions. In a case study, we employ our methods to design a library of dual-target ligands targeting dopamine receptor D2 and 5-hydroxytryptamine receptor 1A as new antipsychotics. Experimental results demonstrate that the proposed model can generate novel compounds with high similarity to both bioactive datasets in several structure-based metrics. Our model exhibits a performance comparable to that of various state-of-the-art multi-objective molecule generation models. We envision that this framework will become a generally applicable approach for designing dual-target drugs in silico.

DGL-LifeSci: An Open-Source Toolkit for Deep Learning on Graphs in Life Science

Graph neural networks (GNNs) constitute a class of deep learning methods for graph data. They have wide applications in chemistry and biology, such as molecular property prediction, reaction prediction, and drug-target interaction prediction. Despite the interest, GNN-based modeling is challenging as it requires graph data preprocessing and modeling in addition to programming and deep learning. Here, we present Deep Graph Library (DGL)-LifeSci, an open-source package for deep learning on graphs in life science. Deep Graph Library (DGL)-LifeSci is a python toolkit based on RDKit, PyTorch, and Deep Graph Library (DGL). DGL-LifeSci allows GNN-based modeling on custom datasets for molecular property prediction, reaction prediction, and molecule generation. With its command-line interfaces, users can perform modeling without any background in programming and deep learning. We test the command-line interfaces using standard benchmarks MoleculeNet, USPTO, and ZINC. Compared with previous implementations, DGL-LifeSci achieves a speed up by up to 6×. For modeling flexibility, DGL-LifeSci provides well-optimized modules for various stages of the modeling pipeline. In addition, DGL-LifeSci provides pretrained models for reproducing the test experiment results and applying models without training. The code is distributed under an Apache-2.0 License and is freely accessible at https://github.com/awslabs/dgl-lifesci.

The Use of Informer Sets in Screening: Perspectives on an Efficient Strategy to Identify New Probes

Small-molecule discovery typically involves large-scale screening campaigns, spanning multiple compound collections. However, such activities can be cost- or time-prohibitive, especially when using complex assay systems, limiting the number of compounds tested. Further, low hit rates can make the process inefficient. Sparse coverage of chemical structure or biological activity space can lead to limited success in a primary screen and represents a missed opportunity by virtue of selecting the "wrong" compounds to test. Thus, the choice of screening collections becomes of paramount importance. In this perspective, we discuss the utility of generating "informer sets" for small-molecule discovery, and how this strategy can be leveraged to prioritize probe candidates. While many researchers may assume that informer sets are focused on particular targets (e.g., kinases) or processes (e.g., autophagy), efforts to assemble informer sets based on historical bioactivity or successful human exposure (e.g., repurposing collections) have shown promise as well. Another method for generating informer sets is based on chemical structure, particularly when the compounds have unknown activities and targets. We describe our efforts to screen an informer set representing a collection of 100,000 small molecules synthesized through diversity-oriented synthesis (DOS). This process enables researchers to identify activity early and more extensively screen only a few chemical scaffolds, rather than the entire collection. This elegant and economic outcome is a goal of the informer set approach. Here, we aim not only to shed light on this process, but also to promote the use of informer sets more widely in small-molecule discovery projects.

FragNet, a Contrastive Learning-Based Transformer Model for Clustering, Interpreting, Visualizing, and Navigating Chemical Space

The question of molecular similarity is core in cheminformatics and is usually assessed via a pairwise comparison based on vectors of properties or molecular fingerprints. We recently exploited variational autoencoders to embed 6M molecules in a chemical space, such that their (Euclidean) distance within the latent space so formed could be assessed within the framework of the entire molecular set. However, the standard objective function used did not seek to manipulate the latent space so as to cluster the molecules based on any perceived similarity. Using a set of some 160,000 molecules of biological relevance, we here bring together three modern elements of deep learning to create a novel and disentangled latent space, viz transformers, contrastive learning, and an embedded autoencoder. The effective dimensionality of the latent space was varied such that clear separation of individual types of molecules could be observed within individual dimensions of the latent space. The capacity of the network was such that many dimensions were not populated at all. As before, we assessed the utility of the representation by comparing clozapine with its near neighbors, and we also did the same for various antibiotics related to flucloxacillin. Transformers, especially when as here coupled with contrastive learning, effectively provide one-shot learning and lead to a successful and disentangled representation of molecular latent spaces that at once uses the entire training set in their construction while allowing "similar" molecules to cluster together in an effective and interpretable way.

A palette of fluorophores that are differentially accumulated by wild-type and mutant strains of Escherichia coli: surrogate ligands for profiling bacterial membrane transporters

Our previous work demonstrated that two commonly used fluorescent dyes that were accumulated by wild-type Escherichia coli MG1655 were differentially transported in single-gene knockout strains, and also that they might be used as surrogates in flow cytometric transporter assays. We summarize the desirable properties of such stains, and here survey 143 candidate dyes. We eventually triage them (on the basis of signal, accumulation levels and cost) to a palette of 39 commercially available and affordable fluorophores that are accumulated significantly by wild-type cells of the 'Keio' strain BW25113, as measured flow cytometrically. Cheminformatic analyses indicate both their similarities and their (much more considerable) structural differences. We describe the effects of pH and of the efflux pump inhibitor chlorpromazine on the accumulation of the dyes. Even the 'wild-type' MG1655 and BW25113 strains can differ significantly in their ability to take up such dyes. We illustrate the highly differential uptake of our dyes into strains with particular lesions in, or overexpressed levels of, three particular transporters or transporter components (yhjV, yihN and tolC). The relatively small collection of dyes described offers a rapid, inexpensive, convenient and informative approach to the assessment of microbial physiology and phenotyping of membrane transporter function.

Microbial natural product databases: moving forward in the multi-omics era

Covering: 2010-2020The digital revolution is driving significant changes in how people store, distribute, and use information. With the advent of new technologies around linked data, machine learning and large-scale network inference, the natural products research field is beginning to embrace real-time sharing and large-scale analysis of digitized experimental data. Databases play a key role in this, as they allow systematic annotation and storage of data for both basic and advanced applications. The quality of the content, structure, and accessibility of these databases all contribute to their usefulness for the scientific community in practice. This review covers the development of databases relevant for microbial natural product discovery during the past decade (2010-2020), including repositories of chemical structures/properties, metabolomics, and genomic data (biosynthetic gene clusters). It provides an overview of the most important databases and their functionalities, highlights some early meta-analyses using such databases, and discusses basic principles to enable widespread interoperability between databases. Furthermore, it points out conceptual and practical challenges in the curation and usage of natural products databases. Finally, the review closes with a discussion of key action points required for the field moving forward, not only for database developers but for any scientist active in the field.

Structural Similarities between Some Common Fluorophores Used in Biology, Marketed Drugs, Endogenous Metabolites, and Natural Products

It is known that at least some fluorophores can act as 'surrogate' substrates for solute carriers (SLCs) involved in pharmaceutical drug uptake, and this promiscuity is taken to reflect at least a certain structural similarity. As part of a comprehensive study seeking the 'natural' substrates of 'orphan' transporters that also serve to take up pharmaceutical drugs into cells, we have noted that many drugs bear structural similarities to natural products. A cursory inspection of common fluorophores indicates that they too are surprisingly 'drug-like', and they also enter at least some cells. Some are also known to be substrates of efflux transporters. Consequently, we sought to assess the structural similarity of common fluorophores to marketed drugs, endogenous mammalian metabolites, and natural products. We used a set of some 150 fluorophores along with standard fingerprinting methods and the Tanimoto similarity metric. Results: The great majority of fluorophores tested exhibited significant similarity (Tanimoto similarity > 0.75) to at least one drug, as judged via descriptor properties (especially their aromaticity, for identifiable reasons that we explain), by molecular fingerprints, by visual inspection, and via the "quantitative estimate of drug likeness" technique. It is concluded that this set of fluorophores does overlap with a significant part of both the drug space and natural products space. Consequently, fluorophores do indeed offer a much wider opportunity than had possibly been realised to be used as surrogate uptake molecules in the competitive or trans-stimulation assay of membrane transporter activities.

DeepGraphMolGen, a multi-objective, computational strategy for generating molecules with desirable properties: a graph convolution and reinforcement learning approach

We address the problem of generating novel molecules with desired interaction properties as a multi-objective optimization problem. Interaction binding models are learned from binding data using graph convolution networks (GCNs). Since the experimentally obtained property scores are recognised as having potentially gross errors, we adopted a robust loss for the model. Combinations of these terms, including drug likeness and synthetic accessibility, are then optimized using reinforcement learning based on a graph convolution policy approach. Some of the molecules generated, while legitimate chemically, can have excellent drug-likeness scores but appear unusual. We provide an example based on the binding potency of small molecules to dopamine transporters. We extend our method successfully to use a multi-objective reward function, in this case for generating novel molecules that bind with dopamine transporters but not with those for norepinephrine. Our method should be generally applicable to the generation in silico of molecules with desirable properties.

VAE-Sim: A Novel Molecular Similarity Measure Based on a Variational Autoencoder

Molecular similarity is an elusive but core "unsupervised" cheminformatics concept, yet different "fingerprint" encodings of molecular structures return very different similarity values, even when using the same similarity metric. Each encoding may be of value when applied to other problems with objective or target functions, implying that a priori none are "better" than the others, nor than encoding-free metrics such as maximum common substructure (MCSS). We here introduce a novel approach to molecular similarity, in the form of a variational autoencoder (VAE). This learns the joint distribution p(z|x) where z is a latent vector and x are the (same) input/output data. It takes the form of a "bowtie"-shaped artificial neural network. In the middle is a "bottleneck layer" or latent vector in which inputs are transformed into, and represented as, a vector of numbers (encoding), with a reverse process (decoding) seeking to return the SMILES string that was the input. We train a VAE on over six million druglike molecules and natural products (including over one million in the final holdout set). The VAE vector distances provide a rapid and novel metric for molecular similarity that is both easily and rapidly calculated. We describe the method and its application to a typical similarity problem in cheminformatics.

Generation of a Small Library of Natural Products Designed to Cover Chemical Space Inexpensively

Natural products space includes at least 200,000 compounds and the structures of most of these compounds are available in digital format. Previous analyses showed (i) that although they were capable of taking up synthetic pharmaceutical drugs, such exogenous molecules were likely the chief ‘natural’ substrates in the evolution of the transporters used to gain cellular entry by pharmaceutical drugs, and (ii) that a relatively simple but rapid clustering algorithm could produce clusters from which individual elements might serve to form a representative library covering natural products space. This exploited the fact that the larger clusters were likely to be formed early in evolution (and hence to have been accompanied by suitable transporters), so that very small clusters, including singletons, could be ignored. In the latter work, we assumed that the molecule chosen might be that in the middle of the cluster. However, this ignored two other criteria, namely the commercial availability and the financial cost of the individual elements of these clusters. We here develop a small representative library in which we to seek to satisfy the somewhat competing criteria of coverage (‘representativeness’), availability and cost. It is intended that the library chosen might serve as a testbed of molecules that may or may not be substrates for known or orphan drug transporters. A supplementary spreadsheet provides details, and their availability via a particular supplier.

Mass Spectrometry-Based Microbial Metabolomics: Techniques, Analysis, and Applications

The demand for understanding the roles genes play in biological systems has steered the biosciences into the direction the metabolome, as it closely reflects the metabolic activities within a cell. The importance of the metabolome is further highlighted by its ability to influence the genome, transcriptome, and proteome. Consequently, metabolomic information is being used to understand microbial metabolic networks. At the forefront of this work is mass spectrometry, the most popular metabolomics measurement technique. Mass spectrometry-based metabolomic analyses have made significant contributions to microbiological research in the environment and human disease. In this chapter, we break down the technical aspects of mass spectrometry-based metabolomics and discuss its application to microbiological research.

Online structure-based screening of purchasable approved drugs and natural compounds: retrospective examples of drug repositioning on cancer targets

Drug discovery is a long and difficult process that benefits from the integration of virtual screening methods in experimental screening campaigns such as to generate testable hypotheses, accelerate and/or reduce the cost of drug development. Current drug attrition rate is still a major issue in all therapeutic areas and especially in the field of cancer. Drug repositioning as well as the screening of natural compounds constitute promising approaches to accelerate and improve the success rate of drug discovery. We developed three compounds libraries of purchasable compounds: Drugs-lib, FOOD-lib and NP-lib that contain approved drugs, food constituents and natural products, respectively, that are optimized for structure-based virtual screening studies. The three compounds libraries are implemented in the MTiOpenScreen web server that allows users to perform structure-based virtual screening computations on their selected protein targets. The server outputs a list of 1,500 molecules with predicted binding scores that can then be processed further by the users and purchased for experimental validation. To illustrate the potential of our service for drug repositioning endeavours, we selected five recently published drugs that have been repositioned in vitro and/or in vivo on cancer targets. For each drug, we used the MTiOpenScreen service to screen the Drugs-lib collection against the corresponding anti-cancer target and we show that our protocol is able to rank these drugs within the top ranked compounds. This web server should assist the discovery of promising molecules that could benefit patients, with faster development times, and reduced costs and risk.

Software Tools and Approaches for Compound Identification of LC-MS/MS Data in Metabolomics

The annotation of small molecules remains a major challenge in untargeted mass spectrometry-based metabolomics. We here critically discuss structured elucidation approaches and software that are designed to help during the annotation of unknown compounds. Only by elucidating unknown metabolites first is it possible to biologically interpret complex systems, to map compounds to pathways and to create reliable predictive metabolic models for translational and clinical research. These strategies include the construction and quality of tandem mass spectral databases such as the coalition of MassBank repositories and investigations of MS/MS matching confidence. We present in silico fragmentation tools such as MS-FINDER, CFM-ID, MetFrag, ChemDistiller and CSI:FingerID that can annotate compounds from existing structure databases and that have been used in the CASMI (critical assessment of small molecule identification) contests. Furthermore, the use of retention time models from liquid chromatography and the utility of collision cross-section modelling from ion mobility experiments are covered. Workflows and published examples of successfully annotated unknown compounds are included.

GeneGini: Assessment via the Gini Coefficient of Reference "Housekeeping" Genes and Diverse Human Transporter Expression Profiles

The expression levels of SLC or ABC membrane transporter transcripts typically differ 100- to 10,000-fold between different tissues. The Gini coefficient characterizes such inequalities and here is used to describe the distribution of the expression of each transporter among different human tissues and cell lines. Many transporters exhibit extremely high Gini coefficients even for common substrates, indicating considerable specialization consistent with divergent evolution. The expression profiles of SLC transporters in different cell lines behave similarly, although Gini coefficients for ABC transporters tend to be larger in cell lines than in tissues, implying selection. Transporter genes are significantly more heterogeneously expressed than the members of most non-transporter gene classes. Transcripts with the stablest expression have a low Gini index and often differ significantly from the "housekeeping" genes commonly used for normalization in transcriptomics/qPCR studies. PCBP1 has a low Gini coefficient, is reasonably expressed, and is an excellent novel reference gene. The approach, referred to as GeneGini, provides rapid and simple characterization of expression-profile distributions and improved normalization of genome-wide expression-profiling data.