Areas Of Chemical Space Research Articles

Comparison of dimensionality reduction techniques for the visualisation of chemical space in organometallic catalysis

We have used a Ligand Knowledge Base for bidentate P,P-donor ligands of potential interest to homogeneous catalysis to compare three dimensionality reduction techniques, namely Principal Component Analysis (PCA), Uniform Manifold Approximation and Projection (UMAP) and t-distributed Stochastic Neighbor Embedding (t-SNE). While our previous work on Ligand Knowledge Bases has focused on PCA, here we compare this approach with more recently-published approaches and assess the information retention, visualization, clustering and interpretability which can be achieved for each approach. We find that potential advantages of t-SNE are not realized with a database of the current size (275 entries), and that there is a degree of complementarity between PCA and UMAP. The statistics underlying PCA rely on linear relationships, making interpretation of the resulting plots comparatively straightforward. Since much of chemistry relies on linear structure-property relationships and low-dimensional visualization, the explainability and information retention achieved is attractive. UMAP proved more challenging to interpret, but achieved clear clustering which was often chemically meaningful, and it would be a useful approach for ensuring that distinct subsets of compounds are sampled in a machine-learning context. This analysis also highlighted that the tunability of catalysis achieved through ligand exchange maps well onto some areas of chemical space where closely related ligands cluster, while others represent outliers; these arise from different combinations of steric and electronic effects which chemists will find intuitive.

Exploring Novel Synthetic Concepts and Strategies Using Mechanochemistry

Abstract In the field of organic synthesis, a ball-milling synthetic technique has garnered significant attention in recent years as an eco-friendly and sustainable alternative to traditional solution-based methods. In addition to its environmental benefits, solid-state organic synthesis using mechanochemical protocols enables access to novel areas of chemical space that are unavailable by conventional solution-based reactions. In this context, we are interested in designing and developing new organic transformations based on the unique solid-state reaction environment and the use of mechanical forces in a ball mill. In this Account, we highlight our latest findings concerning the creation of novel synthetic concepts and strategies. These approaches harness the distinctive reaction environment of mechanochemistry, rather than merely transferring well-established reactions from solution-based protocols to mechanochemical conditions.

Improvements to in silico skin sensitisation predictions through privacy-preserving data sharing

In silico models are often built solely on publicly available data which may mean that they are less predictive for proprietary chemical space. Data sharing initiatives can improve the performance of such models, but organisations are often unable to share their data due to the need to protect their business interests and maintain the confidentiality of the chemicals in their research and development programmes. In silico models like Derek Nexus, which use expert knowledge to develop structural alerts based on chemical toxicity, can use proprietary data to identify new areas of chemical space and/or refine existing alerts whilst still preserving the privacy of the confidential data. Five hundred and thirty seven proprietary chemicals with skin sensitisation data were shared which led to the implementation of 7 new alerts and 5 modified alerts, with a concomitant 19% increase in sensitivity and 3% increase in specificity of the model.

Harnessing Protein-Ligand Interaction Fingerprints to Predict New Scaffolds of RIPK1 Inhibitors.

Necroptosis has emerged as an exciting target in oncological, inflammatory, neurodegenerative, and autoimmune diseases, in addition to acute ischemic injuries. It is known to play a role in innate immune response, as well as in antiviral cellular response. Here we devised a concerted in silico and experimental framework to identify novel RIPK1 inhibitors, a key necroptosis factor. We propose the first in silico model for the prediction of new RIPK1 inhibitor scaffolds by combining docking and machine learning methodologies. Through the data analysis of patterns in docking results, we derived two rules, where rule #1 consisted of a four-residue signature filter, and rule #2 consisted of a six-residue similarity filter based on docking calculations. These were used in consensus with a machine learning QSAR model from data collated from ChEMBL, the literature, in patents, and from PubChem data. The models allowed for good prediction of actives of >90, 92, and 96.4% precision, respectively. As a proof-of-concept, we selected 50 compounds from the ChemBridge database, using a consensus of both molecular docking and machine learning methods, and tested them in a phenotypic necroptosis assay and a biochemical RIPK1 inhibition assay. A total of 7 of the 47 tested compounds demonstrated around 20–25% inhibition of RIPK1’s kinase activity but, more importantly, these compounds were discovered to occupy new areas of chemical space. Although no strong actives were found, they could be candidates for further optimization, particularly because they have new scaffolds. In conclusion, this screening method may prove valuable for future screening efforts as it allows for the exploration of new areas of the chemical space in a very fast and inexpensive manner, therefore providing efficient starting points amenable to further hit-optimization campaigns.

Selecting molecules with diverse structures and properties by maximizing submodular functions of descriptors learned with graph neural networks

Selecting diverse molecules from unexplored areas of chemical space is one of the most important tasks for discovering novel molecules and reactions. This paper proposes a new approach for selecting a subset of diverse molecules from a given molecular list by using two existing techniques studied in machine learning and mathematical optimization: graph neural networks (GNNs) for learning vector representation of molecules and a diverse-selection framework called submodular function maximization. Our method, called SubMo-GNN, first trains a GNN with property prediction tasks, and then the trained GNN transforms molecular graphs into molecular vectors, which capture both properties and structures of molecules. Finally, to obtain a subset of diverse molecules, we define a submodular function, which quantifies the diversity of molecular vectors, and find a subset of molecular vectors with a large submodular function value. This can be done efficiently by using the greedy algorithm, and the diversity of selected molecules measured by the submodular function value is mathematically guaranteed to be at least 63% of that of an optimal selection. We also introduce a new evaluation criterion to measure the diversity of selected molecules based on molecular properties. Computational experiments confirm that our SubMo-GNN successfully selects diverse molecules from the QM9 dataset regarding the property-based criterion, while performing comparably to existing methods regarding standard structure-based criteria. We also demonstrate that SubMo-GNN with a GNN trained on the QM9 dataset can select diverse molecules even from other MoleculeNet datasets whose domains are different from the QM9 dataset. The proposed method enables researchers to obtain diverse sets of molecules for discovering new molecules and novel chemical reactions, and the proposed diversity criterion is useful for discussing the diversity of molecular libraries from a new property-based perspective.

Efficient Exploration of Chemical Space with Docking and Deep Learning.

With the advent of make-on-demand commercial libraries, the number of purchasable compounds available for virtual screening and assay has grown explosively in recent years, with several libraries eclipsing one billion compounds. Today's screening libraries are larger and more diverse, enabling the discovery of more-potent hit compounds and unlocking new areas of chemical space, represented by new core scaffolds. Applying physics-based in silico screening methods in an exhaustive manner, where every molecule in the library must be enumerated and evaluated independently, is increasingly cost-prohibitive. Here, we introduce a protocol for machine learning-enhanced molecular docking based on active learning to dramatically increase throughput over traditional docking. We leverage a novel selection protocol that strikes a balance between two objectives: (1) identifying the best scoring compounds and (2) exploring a large region of chemical space, demonstrating superior performance compared to a purely greedy approach. Together with automated redocking of the top compounds, this method captures almost all the high scoring scaffolds in the library found by exhaustive docking. This protocol is applied to our recent virtual screening campaigns against the D4 and AMPC targets that produced dozens of highly potent, novel inhibitors, and a blind test against the MT1 target. Our protocol recovers more than 80% of the experimentally confirmed hits with a 14-fold reduction in compute cost, and more than 90% of the hit scaffolds in the top 5% of model predictions, preserving the diversity of the experimentally confirmed hit compounds.

Structural‐Based Drug Discovery Targeting PCNA: A Novel Cancer Therapeutic

As a member of the sliding clamp family, the Proliferating Cell Nuclear Antigen (PCNA) critically functions as a hub for proteins involved in DNA metabolism events in close proximity to DNA. Recently, a unique isoform of PCNA has been defined as playing a critical pathological role in various cancers, which includes breast cancer. Notably, this cancer associated isoform of PCNA (caPCNA) is highly expressed in cancer cells, but not at significant levels in non-malignant cells. Recently, our collaborators at the City of Hope Medical Center have identified and developed a set of “first-in-class” small molecules that inhibit caPCNA hub activity in vitro, in cellulo and in vivo, which drive cancer cell death while not affecting non-malignant cells. However, these City of Hope computer aided-drug design (CADD)-based molecules are highly hydrophobic in nature, potentially limiting their bioavailability. We hypothesize that structure-based drug design (SBDD), with a focus on fragment-based drug discovery (FBDD), will allow for the identification of functional groups that can be used to enhance the City of Hope leads and for the development of novel drug-like inhibitors in different areas of chemical space. We have screened over 3,550 drug-like fragments and artificial intelligence (AI)-CADD generated binders revealing 246 promising hits that increase protein thermal stability 2°C or greater, indicating potential binding interaction. Utilizing a micro-seeding technique, we have developed a stable crystallization condition of caPCNA diffracting to 2.4 Å, which will be used to characterize ligand:caPCNA interactions via soaking and co-crystallization studies. Characterization of these binding interactions complemented by activity assays, will enable subsequent SBDD studies to develop hit molecules with improved solubility and stability for lead optimization.

From desktop to benchtop with automated computational workflows for computer-aided design in asymmetric catalysis

The organic chemist’s toolbox is vast, with technologies to accelerate the synthesis of novel chemical matter. The field of asymmetric catalysis is one approach to accessing new areas of chemical space and computational power is today sufficient to assist in this exploration. Unfortunately, existing techniques generally require computational expertise and are therefore underutilized in synthetic chemistry. Here we present our platform Virtual Chemist, which allows bench chemists to predict outcomes of asymmetric chemical reactions ahead of testing in the laboratory, in just a few clicks. Modular workflows facilitate the simulation of various sets of experiments, including the four realistic scenarios discussed: one-by-one design, library screening, hit optimization and substrate-scope evaluation. Catalyst candidates are screened within hours and the enantioselectivity predictions provide substantial enrichments compared to random testing. The achieved accuracies within ~1 kcal mol–1 provide opportunities for computational chemistry in the field of asymmetric catalyst design, allowing bench chemists to guide the design and discovery of asymmetric catalysts. Computational chemistry has remained largely inaccessible to the experimental chemistry community. Here we report the VIRTUAL CHEMIST, a software suite free for academic use, that enables organic chemists without expertise in computational chemistry to perform virtual screening experiments for asymmetric catalyst discovery and design.

Podcast: When rocket chemistry blasted off and came back to Earth

Rocket propellant research had its heyday in the mid-20th century, when the space race and the Cold War meant chemists had plenty of money and long leashes. Only a few of their most interesting ideas ended up in working rockets, but they charted new areas of chemical space, some of which, like boron chemistry, have proved useful in other fields. Geopolitical shifts, along with a growing emphasis on health, safety, and the environment, put a damper on propellant chemistry in the last decades of the 1900s. But the need for high-performance propellants hasn’t gone away, and neither has chemists’ interest in pushing the envelope. In this episode of Stereo Chemistry, we hear from chemists who lived through the heady days of the ’50s and ’60s and the ones carrying rocket chemistry’s torch today. Listen at cenm.ag/rocketfuel.

Tetrahydroindoles as Multipurpose Screening Compounds and Novel Sirtuin Inhibitors.

Indoles are privileged structures in medicinal and bioorganic chemistry that are particularly well suited to serve as platforms for diversity. Among many other therapeutic areas, the indole scaffold has been used to design aromatic compounds useful to interfere with enzymes engaged in the regulation of substrate acylation status, such as sirtuins. However, the planarity of the indole ring is not necessarily optimal for all target enzymes, especially when functionalization with aromatic side chains is required. Replacement of flat scaffolds by nonplanar molecular cores dominated by sp3 hybridization is a common strategy to avoid the disadvantages associated with poor solubility and high promiscuity, while covering less-well-explored areas of chemical space. Thus, we synthesized fragment-like tetrahydroindoles suitable for fragment-based drug discovery as well as a well-characterized small library intended as multipurpose screening compounds. For proof of principle, these compounds were screened against sirtuins 1-3, enzymes known to be addressable by indoles. We found that 2,6,6-trimethyl-4-oxo-4,5,6,7-tetrahydro-1H-indole-3-carboxamides are potent and selective SIRT2 inhibitors. Compound 16 t displayed an IC50 value of 0.98 μm and could serve as exquisite starting point for hit-to-lead profiling.

An Apparent Binary Choice in Biochemistry: Mutual Reactivity Implies Life Chooses Thiols or Nitrogen-Sulfur Bonds, but Not Both.

A fundamental goal of biology is to understand the rules behind life's use of chemical space. Established work focuses on why life uses the chemistry that it does. Given the enormous scope of possible chemical space, we postulate that it is equally important to ask why life largely avoids certain areas of chemical space. The nitrogen-sulfur bond is a prime example, as it rarely appears in natural molecules, despite the very rich N-S bond chemistry applied in various branches of industry (e.g., industrial materials, agrochemicals, pharmaceuticals). We find that, out of more than 200,000 known, unique compounds made by life, only about 100 contain N-S bonds. Furthermore, the limited number of N-S bond-containing molecules that life produces appears to fall into a few very distinctive structural groups. One may think that industrial processes are unrelated to biochemistry because of a greater possibility of solvents, catalysts, and temperatures available to industry than to the cellular environment. However, the fact that life does rarely make N-S bonds, from the plentiful precursors available, and has evolved the ability to do so independently several times, suggests that the restriction on life's use of N-S chemistry is not in its synthesis. We present a hypothesis to explain life's extremely limited usage of the N-S bond: that the N-S bond chemistry is incompatible with essential segments of biochemistry, specifically with thiols. We support our hypothesis by (1) a quantitative analysis of the occurrence of N-S bond-containing natural products and (2) reactivity experiments between selected N-S compounds and key biological molecules. This work provides an example of a reason why life nearly excludes a distinct region of chemical space. Combined with future examples, this potentially new field of research may provide fresh insight into life's evolution through chemical space and its origin and early evolution.

Strategies towards expansion of chemical space of natural product-based compounds to enable drug discovery

Natural products (NPs) are an excellent source of biologically active molecules that provide many biologically biased features that enable innovative designing of synthetic compounds. NPs are characterized by high content of sp3 -hybridized carbon atoms; oxygen; spiro, bridged, and linked systems; and stereogenic centers, with high structural diversity. To date, several approaches have been implemented for mapping and navigating into the chemical space of NPs to explore the different aspects of chemical space. The approaches providing novel opportunities to synthesize NP-inspired compound libraries involve NP-based fragments and ring distortion strategies. These methodologies allow access to areas of chemical space that are less explored, and consequently help to overcome the limitations in the use of NPs in drug discovery, such as lack of accessibility and synthetic intractability. In this review, we describe how NPs have recently been used as a platform for the development of diverse compounds with high structural and stereochemical complexity. In addition, we show developed strategies aiming to reengineer NPs toward the expansion of NP-based chemical space by fragment-based approaches and chemical degradation to yield novel compounds to enable drug discovery.

The Rational Discovery of a Tau Aggregation Inhibitor.

Intrinsically disordered proteins play vital roles in biology, and their dysfunction contributes to many major disease states. These proteins remain challenging targets for rational ligand discovery or drug design because they are highly dynamic and fluctuate through a diverse set of conformations, frustrating structure-based approaches. To meet this challenge, we have developed protocols to efficiently identify active small molecule ligands of disordered proteins. Our approach utilizes enhanced sampling molecular dynamics and conformational analysis approaches optimized for disordered targets, coupled with computational docking and machine learning-based screens of compound libraries. By applying this protocol to an amyloid-forming segment of microtubule-associated protein tau, we successfully identified novel, chemically diverse tau ligands, including an inhibitor that delays the aggregation reaction in vitro without affecting the amount of aggregate formed at the steady state. Our results indicate that we have expanded the toolkit of protein aggregation inhibitors into new areas of chemical space and demonstrate the feasibility of our ligand discovery strategy.

Catalyst: The Metaphysics of Chemical Reactivity

Leroy (Lee) Cronin has been thinking about programming chemistry since he was 8 years old, when he got his first computer and chemistry set. He is now the Regius Professor of Chemistry at the University of Glasgow. He and his team are trying to make artificial life forms, find alien life, explore the digitization of chemistry, and construct chemical computers. S. Hessam M. Mehr is a post-doctoral researcher at the University of Glasgow. He earned his PhD at the University of British Columbia (2017). His research focuses on probabilistic and information-theoretic methods in chemistry and the discovery of chemical novelty. Jaroslaw M. Granda is a postdoctoral research associate at the University of Glasgow. He obtained his PhD at the Institute of Organic Chemistry of the Polish Academy of Sciences (2014). His research interests focus on the automation of organic synthesis and artificial-intelligence-driven systems for chemical discovery.

Extending in Silico Protein Target Prediction Models to Include Functional Effects.

In silico protein target deconvolution is frequently used for mechanism-of-action investigations; however existing protocols usually do not predict compound functional effects, such as activation or inhibition, upon binding to their protein counterparts. This study is hence concerned with including functional effects in target prediction. To this end, we assimilated a bioactivity training set for 332 targets, comprising 817,239 active data points with unknown functional effect (binding data) and 20,761,260 inactive compounds, along with 226,045 activating and 1,032,439 inhibiting data points from functional screens. Chemical space analysis of the data first showed some separation between compound sets (binding and inhibiting compounds were more similar to each other than both binding and activating or activating and inhibiting compounds), providing a rationale for implementing functional prediction models. We employed three different architectures to predict functional response, ranging from simplistic random forest models (‘Arch1’) to cascaded models which use separate binding and functional effect classification steps (‘Arch2’ and ‘Arch3’), differing in the way training sets were generated. Fivefold stratified cross-validation outlined cascading predictions provides superior precision and recall based on an internal test set. We next prospectively validated the architectures using a temporal set of 153,467 of in-house data points (after a 4-month interim from initial data extraction). Results outlined Arch3 performed with the highest target class averaged precision and recall scores of 71% and 53%, which we attribute to the use of inactive background sets. Distance-based applicability domain (AD) analysis outlined that Arch3 provides superior extrapolation into novel areas of chemical space, and thus based on the results presented here, propose as the most suitable architecture for the functional effect prediction of small molecules. We finally conclude including functional effects could provide vital insight in future studies, to annotate cases of unanticipated functional changeover, as outlined by our CHRM1 case study.

Synthesis of Constrained Heterocycles Employing Two Post-Ugi Cyclization Methods for Rapid Library Generation with In Cellulo Activity.

Benzimidazoles and quinoxalinones are present in the core of many pharmacologically relevant compounds. Several combinatorial methods have been developed to attach ring systems to both scaffolds for derivatization at select positions. Herein, we describe the development of novel constrained heterocyclic compounds attached to the N1 position of both benzimidazole and quinoxalinone scaffolds. Utilizing robust post-Ugi cyclization methods, including the Ugi-deprotection-cyclization (UDC) methodology, allows for efficient access to a new area of chemical space. Additionally, molecular modeling and in cellulo screening was employed to therapeutically validate the compounds formed with this method.

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

Armed with the digital availability of two natural products libraries, amounting to some 195 885 molecular entities, we ask the question of how we can best sample from them to maximize their "representativeness" in smaller and more usable libraries of 96, 384, 1152, and 1920 molecules. The term "representativeness" is intended to include diversity, but for numerical reasons (and the likelihood of being able to perform a QSAR) it is necessary to focus on areas of chemical space that are more highly populated. Encoding chemical structures as fingerprints using the RDKit "patterned" algorithm, we first assess the granularity of the natural products space using a simple clustering algorithm, showing that there are major regions of "denseness" but also a great many very sparsely populated areas. We then apply a "hybrid" hierarchical K-means clustering algorithm to the data to produce more statistically robust clusters from which representative and appropriate numbers of samples may be chosen. There is necessarily again a trade-off between cluster size and cluster number, but within these constraints, libraries containing 384 or 1152 molecules can be found that come from clusters that represent some 18 and 30% of the whole chemical space, with cluster sizes of, respectively, 50 and 27 or above, just about sufficient to perform a QSAR. By using the online availability of molecules via the Molport system (www.molport.com), we are also able to construct (and, for the first time, provide the contents of) a small virtual library of available molecules that provided effective coverage of the chemical space described. Consistent with this, the average molecular similarities of the contents of the libraries developed is considerably smaller than is that of the original libraries. The suggested libraries may have use in molecular or phenotypic screening, including for determining possible transporter substrates.

ChemInform Abstract: Interrogating Novel Areas of Chemical Space for Drug Discovery Using Chemoinformatics

AbstractReview: 87 refs.

Pragmatic Approaches to Using Computational Methods To Predict Xenobiotic Metabolism

In this study the performance of a selection of computational models for the prediction of metabolites and/or sites of metabolism was investigated. These included models incorporated in the MetaPrint2D-React, Meteor, and SMARTCyp software. The algorithms were assessed using two data sets: one a homogeneous data set of 28 Non-Steroidal Anti-Inflammatory Drugs (NSAIDs) and paracetamol (DS1) and the second a diverse data set of 30 top-selling drugs (DS2). The prediction of metabolites for the diverse data set (DS2) was better than for the more homogeneous DS1 for each model, indicating that some areas of chemical space may be better represented than others in the data used to develop and train the models. The study also identified compounds for which none of the packages could predict metabolites, again indicating areas of chemical space where more information is needed. Pragmatic approaches to using metabolism prediction software have also been proposed based on the results described here. These approaches include using cutoff values instead of restrictive reasoning settings in Meteor to reduce the output with little loss of sensitivity and for directing metabolite prediction by preselection based on likely sites of metabolism.

IADE: a system for intelligent automatic design of bioisosteric analogs

IADE, a software system supporting molecular modellers through the automatic design of non-classical bioisosteric analogs, scaffold hopping and fragment growing, is presented. The program combines sophisticated cheminformatics functionalities for constructing novel analogs and filtering them based on their drug-likeness and synthetic accessibility using automatic structure-based design capabilities: the best candidates are selected according to their similarity to the template ligand and to their interactions with the protein binding site. IADE works in an iterative manner, improving the fitness of designed molecules in every generation until structures with optimal properties are identified. The program frees molecular modellers from routine, repetitive tasks, allowing them to focus on analysis and evaluation of the automatically designed analogs, considerably enhancing their work efficiency as well as the area of chemical space that can be covered. The performance of IADE is illustrated through a case study of the design of a nonclassical bioisosteric analog of a farnesyltransferase inhibitor--an analog that has won a recent "Design a Molecule" competition.