Generating a knowledge base of new activity cliffs (ACs) defined on the basis of compound set-dependent potency distributions, also taking confirmed inactive compounds into account.

Target deconvolution of phenotypic assays is a hot topic in chemical biology and drug discovery. The ultimate goal is the identification of targets for compounds that produce interesting phenotypic readouts. A variety of experimental and computational strategies have been devised to aid this process. A widely applied computational approach infers putative targets of new active molecules on the basis of their chemical similarity to compounds with activity against known targets. Herein, we introduce a molecular scaffold-based variant for similarity-based target deconvolution from chemical cancer cell line screens that were used as a model system for phenotypic assays. A new scaffold type was used for substructure-based similarity assessment, termed analog series-based (ASB) scaffold. Compared with conventional scaffolds and compound-based similarity calculations, target assignment centered on ASB scaffolds resulting from screening hits and bioactive reference compounds restricted the number of target hypotheses in a meaningful way and lead to a significant enrichment of known cancer targets among candidates.

Allosteric and ATP-competitive kinase inhibitors act by distinct mechanisms and are expected to have high and low kinase selectivity, respectively. This also raises the question whether or not these different types of inhibitors might be structurally distinct. To address this question, we have assembled data sets of currently available competitive and allosteric kinase inhibitors confirmed by X-ray crystallography and systematically compared these compounds on the basis of different structural criteria. Many competitive and allosteric inhibitors were found to contain the same or similar substructures and a subset of allosteric inhibitors was found to share core structures with ATP site-directed inhibitors. In some instances, small chemical modifications of common cores were found to yield either allosteric or competitive inhibitors. Hence, these different categories of inhibitors with distinct mechanisms of action were often structurally related and represented much more of a structural continuum than discrete states. Additional target annotations were frequently identified for competitive inhibitors, but were rare for allosteric inhibitors. As a part of this study, our collection of kinase inhibitors and the associated information are made freely available to enable further assessment of chemical modifications that distinguish similar kinase inhibitors with distinct mechanisms of action.

Activity cliffs are formed by structurally analogous compounds with large potency variations and are highly relevant for the exploration of discontinuous structure-activity relationships and compound optimization. So far, activity cliffs have mostly been studied on a case-by-case basis or assessed by global statistical analysis. Different from previous investigations, we report a large-scale analysis of activity cliff formation with a strong focus on individual compound activity classes (target sets). Compound potency distributions were systematically analyzed and categorized, and structural relationships were dissected and visualized on a per-set basis. Our study uncovered target set-dependent interplay of potency distributions and structural relationships and revealed the presence of activity cliffs and origins of cliff formation in different structure-activity relationship environments.

Generation of an R-group replacement system for compound optimization in medicinal chemistry.

Predicting compounds with single- and multi-target activity and exploring origins of compound specificity and promiscuity is of high interest for chemical biology and drug discovery. We present a large-scale analysis of compound promiscuity including two major components. First, high-confidence datasets of compounds with multi- and corresponding single-target activity were extracted from biological screening data. Positive and negative assay results were taken into account and data completeness was ensured. Second, these datasets were investigated using diagnostic machine learning to systematically distinguish between compounds with multi- and single-target activity. Models built on the basis of chemical structure consistently produced meaningful predictions. These findings provided evidence for the presence of structural features differentiating promiscuous and non-promiscuous compounds. Machine learning under varying conditions using modified datasets revealed a strong influence of nearest neighbor relationship on the predictions. Many multi-target compounds were found to be more similar to other multi-target compounds than single-target compounds and vice versa, which resulted in consistently accurate predictions. The results of our study confirm the presence of structural relationships that differentiate promiscuous and non-promiscuous compounds.

Providing compound data sets for promiscuity analysis with single-target (ST) and multi-target (MT) activity, taking confirmed inactivity against targets into account.

Activity landscape (AL) models are used for visualizing and interpreting structure-activity relationships (SARs) in compound datasets. Therefore, ALs are designed to present chemical similarity and compound potency information in context. Different two- or three-dimensional (2D or 3D) AL representations have been introduced. For SAR analysis, 3D AL models are particularly intuitive. In these models, an interpolated potency surface is added as a third dimension to a 2D projection of chemical space. Accordingly, AL topology can be associated with characteristic SAR features. Going beyond visualization and a qualitative assessment of SARs, it would be very helpful to compare 3D ALs of different datasets in more quantitative terms. However, quantitative AL analysis is still in its infancy. Recently, it has been shown that 3D AL models with pre-defined topologies can be correctly classified using machine learning. Classification was facilitated on the basis of AL image feature representations learned with convolutional neural networks. Therefore, we have further investigated image analysis for quantitative comparison of 3D ALs and devised an approach to determine (dis)similarity relationships for ALs representing different compound datasets. Herein, we report this approach and demonstrate proof-of-principle. The methodology makes it possible to computationally compare 3D ALs and quantify topological differences reflecting varying SAR information content. For SAR exploration in drug design, this adds a quantitative measure of AL (dis)similarity to graphical analysis.

Compound profiling matrices record assay results for compound libraries tested against panels of targets. In addition to their relevance for exploring structure-activity relationships, such matrices are of considerable interest for chemoinformatic and chemogenomic applications. For example, profiling matrices provide a valuable data resource for the development and evaluation of machine learning approaches for multitask activity prediction. However, experimental compound profiling matrices are rare in the public domain. Although they are generated in pharmaceutical settings, they are typically not disclosed. Herein, we present an algorithm for the generation of large profiling matrices, for example, containing more than 100 000 compounds exhaustively tested against 50 to 100 targets. The new methodology is a variant of biclustering algorithms originally introduced for large-scale analysis of genomics data. Our approach is applied here to assays from the PubChem BioAssay database and generates profiling matrices of increasing assay or compound coverage by iterative removal of entities that limit coverage. Weight settings control final matrix size by preferentially retaining assays or compounds. In addition, the methodology can also be applied to generate matrices enriched with active entries representing above-average assay hit rates.

In medicinal chemistry and chemoinformatics, activity cliffs (ACs) are defined as pairs of structurally similar compounds that are active against the same target but have a large difference in potency. Accordingly, ACs are rich in structure-activity relationship (SAR) information, which rationalizes their relevance for medicinal chemistry. For identifying ACs, a compound similarity criterion and a potency difference criterion must be specified. So far a constant potency difference between AC partner compounds has mostly been set, e.g. 100-fold, irrespective of the specific activity (targets) of cliff-forming compounds. Herein, we introduce a computational methodology for AC identification and analysis that includes three novel components: •ACs are identified on the basis of variable target set-dependent potency difference criteria (a 'target set' represents a collection of compounds that are active against a given target protein).•ACs are extracted from computationally determined analog series (ASs) and consist of pairs of analogs with single or multiple substitution sites.•For multi-site ACs, a search for analogs with individual substitutions is performed to analyze their contributions to AC formation and determine if multi-site ACs can be represented by single-site ACs.

Scaffold hopping and activity cliff formation define opposite ends of the activity landscape feature spectrum. To rationalize these events at the level of scaffolds, active compounds involved in scaffold hopping were required to contain topologically distinct scaffolds but have only limited differences in potency, whereas compounds involved in activity cliffs were required to share the same scaffold but have large differences in potency. A systematic search was carried out for compounds involved in scaffold hopping and/or activity cliff formation. Results obtained for compound data sets covering more than 300 human targets revealed clear trends. If scaffolds represented multiple but fewer than 10 active compounds, nearly 90% of all scaffolds were exclusively involved in hopping events. With increasing compound coverage, the fraction of scaffolds involved in both scaffold hopping and activity cliff formation significantly increased to more than 50%. However, ∼40% of the scaffolds representing large numbers of active compounds continued to be exclusively involved in scaffold hopping. More than 200 scaffolds with broad target coverage were identified that consistently represented potent compounds and yielded an abundance of scaffold hops in the low-nanomolar range. These and other subsets of scaffolds we characterized are of prime interest for structure-activity relationship (SAR) exploration and compound design. Therefore, the complete scaffold classification generated in the course of our analysis is made freely available.

From currently available kinase inhibitors and their activity data, clusters of coordinated activity cliffs were systematically derived and subjected to cluster index and index map analysis. Type I-like inhibitors with well-defined IC50 measurements were found to provide a large knowledge base of activity cliff clusters for 266 targets from nine kinase groups. On the basis of index map analysis, these clusters were systematically organized according to structural similarity of inhibitors and activity cliff diversity and prioritized for structure-activity relationship (SAR) analysis. From prioritized clusters, interpretable SAR information can be extracted. It is also shown that activity cliff clusters formed by ATP site-directed inhibitors often represent local SAR environments of rather different complexity and interpretability. In addition, activity cliff clusters including promiscuous kinase inhibitors have been determined. Only a small subset of inhibitors was found to change activity cliff roles in different clusters. The activity cliff clusters described herein and their index map organization substantially enrich SAR information associated with kinase inhibitors in compound subsets of limited size. The cluster and index map information is made available upon request to provide opportunities for further SAR exploration. On the basis of our analysis and the data provided, activity cliff clusters and corresponding inhibitor series for kinase targets of interest can be readily selected.

Activity cliffs (ACs) are formed by structurally similar active compounds with large potency differences. In medicinal chemistry, ACs are of high interest because they reveal structure-activity relationship (SAR) information and SAR determinants. Herein, we introduce a new type of ACs that consist of analog pairs with different substitutions at multiple sites (multi-site ACs; msACs). A systematic search for msACs across different classes of bioactive compounds identified more than 4000 of such ACs, most of which had substitutions at two sites (dual-site ACs; dsACs). A hierarchical analog data structure was designed to analyze contributions of individual substitutions to AC formation. Single substitutions were frequently found to determine potency differences captured by dsACs. Hence, in such cases, there was redundancy of AC information. In instances where both substitutions made significant contributions to dsACs, additive, synergistic, and compensatory effects were observed. Taken together, the results of our analysis revealed the prevalence of single-site ACs (ssACs) in analog series, followed by dsACs, which reveal different ways in which paired substitutions contribute to the formation of ACs and modulate SARs.

Compounds with defined multi-target activity play an increasingly important role in drug discovery. Structural features that might be signatures of such compounds have mostly remained elusive thus far. We have explored the potential of explainable machine learning to uncover structural motifs that are characteristic of dual-target compounds. For a pharmacologically relevant target pair-based test system designed for our study, accurate prediction models were derived and the influence of molecular representation features of test compounds was quantified to explain the predictions. The analysis revealed small numbers of specific features whose presence in dual-target and absence in single-target compounds determined accurate predictions. These features formed coherent substructures in dual-target compounds. From computational analysis of specific feature contributions, structural motifs emerged that were confirmed to be signatures of different dual-target activities. Our findings demonstrate the ability of explainable machine learning to bridge between predictions and intuitive chemical analysis and reveal characteristic substructures of dual-target compounds.

An activity cliff (AC) is formed by a pair of structurally similar compounds with a large difference in potency. Accordingly, ACs reveal structure-activity relationship (SAR) discontinuity and provide SAR information for compound optimization. Herein, we have investigated the question if ACs could be predicted from image data. Therefore, pairs of structural analogs were extracted from different compound activity classes that formed or did not form ACs. From these compound pairs, consistently formatted images were generated. Image sets were used to train and test convolutional neural network (CNN) models to systematically distinguish between ACs and non-ACs. The CNN models were found to predict ACs with overall high accuracy, as assessed using alternative performance measures, hence establishing proof-of-principle. Moreover, gradient weights from convolutional layers were mapped to test compounds and identified characteristic structural features that contributed to successful predictions. Weight-based feature visualization revealed the ability of CNN models to learn chemistry from images at a high level of resolution and aided in the interpretation of model decisions with intrinsic black box character.

The ability of target proteins to bind structurally diverse compounds and compounds with different degrees of promiscuity (multi-target activity) was systematically assessed on the basis of currently available activity data and target annotations. Intuitive first- and second-order target promiscuity indices were introduced to quantify these binding characteristics and relate them to each other. For compounds and targets, opposite promiscuity trends were observed. Furthermore, the analysis detected many targets that interacted with compounds representing a similar degree of structural diversity but displayed strong tendencies to recognize either promiscuous or selective compounds. Moreover, target families were identified that preferentially interacted with promiscuous compounds. Taken together, these findings further extend our understanding of the molecular basis of polypharmacology.

In compound optimization, analogue series (ASs) are generated by introducing different R-groups (substituents, functional groups) at specific substitution sites. Systematic investigations of R-groups in medicinal chemistry have so far been rare. We have carried out a large-scale computational analysis of R-groups on the basis of ASs covering currently available bioactive compounds (Takeuchi et al., 2021). With the aid of a network data structure, frequently used R-groups and preferred replacements were identified. On the basis of these data, R-group replacement hierarchies were derived and organized in a searchable database that is made freely available. This contribution complements our systematic analysis (Takeuchi et al., 2021) by specifying the data we have generated and detailing their open access deposition.

Compound promiscuity is defined as the ability of a small molecule to specifically interact with multiple biological targets. So-defined promiscuity is relevant for drug discovery because it provides the molecular basis of polypharmacology, which is increasingly implicated in the therapeutic efficacy of drugs. Recent studies have analyzed different aspects of compound promiscuity on the basis of currently available activity data. In this commentary, we present take-home messages from these studies augmented with new results to generate a detailed picture of compound promiscuity that might serve as a reference for further discussions and research activities.

Selection of R-groups (substituents, functional groups) is of critical importance for the generation of analogues during hit-to-lead and lead optimization. In the practice of medicinal chemistry, R-group selection is mostly driven by chemical experience and intuition taking synthetic criteria into account. However, systematic analyses of substituents are currently rare. In this work, we have computationally isolated R-groups from more than 17,000 analog series comprising ∼315,000 bioactive compounds. From more than 50,000 unique substituents, frequently used R-groups were identified. For these R-groups, preferred replacements over more than 60,000 individual substitution sites were identified with the aid of a network data structure. These data provided the basis for the generation of a searchable R-group replacement system for medicinal chemistry containing replacement hierarchies for frequently used R-groups, which is made freely available as the central component of our study.

In 2012, we reported 30 compound data sets and/or programs developed in our laboratory in a data article and made them freely available to the scientific community to support chemoinformatics and computational medicinal chemistry applications. These data sets and computational tools were provided for download from our website. Since publication of this data article, we have generated 13 new data sets with which we further extend our collection of publicly available data and tools. Due to changes in web servers and website architectures, data accessibility has recently been limited at times. Therefore, we have also transferred our data sets and tools to a public repository to ensure full and stable accessibility. To aid in data selection, we have classified the data sets according to scientific subject areas. Herein, we describe new data sets, introduce the data organization scheme, summarize the database content and provide detailed access information in ZENODO (doi: 10.5281/zenodo.8451 and doi:10.5281/zenodo.8455).