Design of chemical space networks on the basis of Tversky similarity.

Chemical space networks (CSNs) have been introduced as a coordinate-free representation of chemical space. In CSNs, nodes represent compounds and edges pairwise similarity relationships. These network representations are mostly used to navigate sections of biologically relevant chemical space. Different types of CSNs have been designed on the basis of alternative similarity measures including continuous numerical similarity values or substructure-based similarity criteria. CSNs can be characterized and compared on the basis of statistical concepts from network science. Herein, a new CSN design is introduced that is based upon asymmetric similarity assessment using the Tversky coefficient and termed TV-CSN. Compared to other CSNs, TV-CSNs have unique features. While CSNs typically contain separate compound communities and exhibit small world character, many TV-CSNs are also scale-free in nature and contain hubs, i.e., extensively connected central compounds. Compared to other CSNs, these hubs are a characteristic of TV-CSN topology. Hub-containing compound communities are of particular interest for the exploration of structure-activity relationships.

Lessons learned from the design of chemical space networks and opportunities for new applications.

The concept of chemical space is of fundamental relevance in chemical informatics and computer-aided drug discovery. In a series of articles published in the Journal of Computer-Aided Molecular Design, principles of chemical space design were evaluated, molecular networks proposed as an alternative to conventional coordinate-based chemical reference spaces, and different types of chemical space networks (CSNs) constructed and analyzed. Central to the generation of CSNs was the way in which molecular similarity relationships were assessed and a primary focal point was the network-based representation of biologically relevant chemical space. The design and comparison of CSNs based upon alternative similarity measures can be viewed as an evolutionary path with interesting lessons learned along the way. CSN design has matured to the point that such chemical space representations can be used in practice. In this contribution, highlights from the sequence of CSN design efforts are discussed in context, providing a perspective for future practical applications.

Design of chemical space networks using a Tanimoto similarity variant based upon maximum common substructures.

Chemical space networks (CSNs) have recently been introduced as an alternative to other coordinate-free and coordinate-based chemical space representations. In CSNs, nodes represent compounds and edges pairwise similarity relationships. In addition, nodes are annotated with compound property information such as biological activity. CSNs have been applied to view biologically relevant chemical space in comparison to random chemical space samples and found to display well-resolved topologies at low edge density levels. The way in which molecular similarity relationships are assessed is an important determinant of CSN topology. Previous CSN versions were based on numerical similarity functions or the assessment of substructure-based similarity. Herein, we report a new CSN design that is based upon combined numerical and substructure similarity evaluation. This has been facilitated by calculating numerical similarity values on the basis of maximum common substructures (MCSs) of compounds, leading to the introduction of MCS-based CSNs (MCS-CSNs). This CSN design combines advantages of continuous numerical similarity functions with a robust and chemically intuitive substructure-based assessment. Compared to earlier version of CSNs, MCS-CSNs are characterized by a further improved organization of local compound communities as exemplified by the delineation of drug-like subspaces in regions of biologically relevant chemical space.

Design and characterization of chemical space networks for different compound data sets.

Chemical Space Networks (CSNs) are generated for different compound data sets on the basis of pairwise similarity relationships. Such networks are thought to complement and further extend traditional coordinate-based views of chemical space. Our proof-of-concept study focuses on CSNs based upon fingerprint similarity relationships calculated using the conventional Tanimoto similarity metric. The resulting CSNs are characterized with statistical measures from network science and compared in different ways. We show that the homophily principle, which is widely considered in the context of social networks, is a major determinant of the topology of CSNs of bioactive compounds, designed as threshold networks, typically giving rise to community structures. Many properties of CSNs are influenced by numerical features of the conventional Tanimoto similarity metric and largely dominated by the edge density of the networks, which depends on chosen similarity threshold values. However, properties of different CSNs with constant edge density can be directly compared, revealing systematic differences between CSNs generated from randomly collected or bioactive compounds.

Comparison of bioactive chemical space networks generated using substructure- and fingerprint-based measures of molecular similarity.

Chemical space networks (CSNs) have recently been introduced as a conceptual alternative to coordinate-based representations of chemical space. CSNs were initially designed as threshold networks using the Tanimoto coefficient as a continuous similarity measure. The analysis of CSNs generated from sets of bioactive compounds revealed that many statistical properties were strongly dependent on their edge density. While it was difficult to compare CSNs at pre-defined similarity threshold values, CSNs with constant edge density were directly comparable. In the current study, alternative CSN representations were constructed by applying the matched molecular pair (MMP) formalism as a substructure-based similarity criterion. For more than 150 compound activity classes, MMP-based CSNs (MMP-CSNs) were compared to corresponding threshold CSNs (THR-CSNs) at a constant edge density by applying different parameters from network science, measures of community structure distributions, and indicators of structure-activity relationship (SAR) information content. MMP-CSNs were found to be an attractive alternative to THR-CSNs, yielding low edge densities and well-resolved topologies. MMP-CSNs and corresponding THR-CSNs often had similar topology and closely corresponding community structures, although there was only limited overlap in similarity relationships. The homophily principle from network science was shown to affect MMP-CSNs and THR-CSNs in different ways, despite the presence of conserved topological features. Moreover, activity cliff distributions in alternative CSN designs markedly differed, which has important implications for SAR analysis.

Chemical space networks: a powerful new paradigm for the description of chemical space.

The concept of chemical space is playing an increasingly important role in many areas of chemical research, especially medicinal chemistry and chemical biology. It is generally conceived as consisting of numerous compound clusters of varying sizes scattered throughout the space in much the same way as galaxies of stars inhabit our universe. A number of issues associated with this coordinate-based representation are discussed. Not the least of which is the continuous nature of the space, a feature not entirely compatible with the inherently discrete nature of chemical space. Cell-based representations, which are derived from coordinate-based spaces, have also been developed that facilitate a number of chemical informatic activities (e.g., diverse subset selection, filling 'diversity voids', and comparing compound collections).These representations generally suffer the 'curse of dimensionality'. In this work, networks are proposed as an attractive paradigm for representing chemical space since they circumvent many of the issues associated with coordinate- and cell-based representations, including the curse of dimensionality. In addition, their relational structure is entirely compatible with the intrinsic nature of chemical space. A description of the features of these chemical space networks is presented that emphasizes their statistical characteristics and indicates how they are related to various types of network topologies that exhibit random, scale-free, and/or 'small world' properties.

Conditional probabilities of activity landscape features for individual compounds.

Activity landscape representations aid in the analysis of structure-activity relationships (SARs) of large compound data sets. Landscapes are characterized by features with different SAR information content such as, for example, regions formed by structurally diverse compounds having similar activity or, alternatively, structurally similar compounds with large activity differences, so-called activity cliffs. Modeling of activity landscapes typically requires pairwise comparisons of molecular similarity and potency relationships of compounds in a data set. Consequently, landscape features are generally resolved at the level of compound pairs. Herein, we introduce a methodology to assign feature probabilities to individual compounds. This makes it possible to organize compounds comprising activity landscapes into well-defined SAR categories. Specifically, the calculation of conditional feature probabilities of active compounds provides a balanced and further refined view of activity landscapes with a focus on individual molecules.

Activity cliffs in PubChem confirmatory bioassays taking inactive compounds into account.

Activity cliffs are formed by pairs or groups of structurally similar compounds with significant differences in potency. They represent a prominent feature of activity landscapes of compound data sets and a primary source of structure-activity relationship (SAR) information. Thus far, activity cliffs have only been considered for active compounds, consistent with the principles of the activity landscape concept. However, from an SAR perspective, pairs formed by structurally similar active and inactive compounds should often also be informative. Therefore, we have extended the activity cliff concept to also take inactive compounds into consideration. As source of both confirmed active and inactive compounds, we have exclusively focused on PubChem confirmatory bioassays. Activity cliffs formed between pairs of active compounds (homogeneous pairs) and pairs of active and inactive compounds (heterogeneous pairs) were systematically analyzed on a per-assay basis, hence ensuring the currently highest possible degree of experimental consistency in activity measurement. Only very small numbers of large-magnitude activity cliffs formed between active compounds were detected in PubChem bioassays. However, when taking confirmed inactive compounds from confirmatory assays into account, the activity cliff frequency in assay data significantly increased, involving 11-15% of all qualifying pairs of similar compounds, depending on the molecular representations that were used. Hence, these non-conventional activity cliffs provide an additional source of SAR information.

Is there a future for computational chemistry in drug research?

Improvements in computational chemistry methods have had a growing impact on drug research. But will incremental improvements be sufficient to ensure this continues? Almost all existing efforts to discover new drugs depend on the classic one target, one drug paradigm, although the situation is changing slowly. A new paradigm that focuses on a more systems biology approach and takes account of the reality that most drugs exhibit some level of polypharmacology is beginning to emerge. This will bring about dramatic changes that can significantly influence the role that computational methods play in future drug research. But these changes require that current methods be augmented with those from bioinformatics and engineering if the field is to have a significant impact on future drug research.

Chemoinformatics and artificial intelligence colloquium: progress and challenges in developing bioactive compounds.

We report the main conclusions of the first Chemoinformatics and Artificial Intelligence Colloquium, Mexico City, June 15-17, 2022. Fifteen lectures were presented during a virtual public event with speakers from industry, academia, and non-for-profit organizations. Twelve hundred and ninety students and academics from more than 60 countries. During the meeting, applications, challenges, and opportunities in drug discovery, de novo drug design, ADME-Tox (absorption, distribution, metabolism, excretion and toxicity) property predictions, organic chemistry, peptides, and antibiotic resistance were discussed. The program along with the recordings of all sessions are freely available at https://www.difacquim.com/english/events/2022-colloquium/ .

Consensus models of activity landscapes with multiple chemical, conformer, and property representations.

We report consensus Structure-Activity Similarity (SAS) maps that address the dependence of activity landscapes on molecular representation. As a case study, we characterized the activity landscape of 54 compounds with activities against human cathepsin B (hCatB), human cathepsin L (hCatL), and Trypanosoma brucei cathepsin B (TbCatB). Starting from an initial set of 28 descriptors we selected ten representations that capture different aspects of the chemical structures. These included four 2D (MACCS keys, GpiDAPH3, pairwise, and radial fingerprints) and six 3D (4p and piDAPH4 fingerprints with each including three conformers) representations. Multiple conformers are used for the first time in consensus activity landscape modeling. The results emphasize the feasibility of identifying consensus data points that are consistently formed in different reference spaces generated with several fingerprint models, including multiple 3D conformers. Consensus data points are not meant to eliminate data, disregarding, for example, "true" activity cliffs that are not identified by some molecular representations. Instead, consensus models are designed to prioritize the SAR analysis of activity cliffs and other consistent regions in the activity landscape that are captured by several molecular representations. Systematic description of the SARs of two targets give rise to the identification of pairs of compounds located in the same region of the activity landscape of hCatL and TbCatB suggesting similar mechanisms of action for the pairs involved. We also explored the relationship between property similarity and activity similarity and found that property similarities are suitable to characterize SARs. We also introduce the concept of structure-property-activity (SPA) similarity in SAR studies.

The reductionist paradox: are the laws of chemistry and physics sufficient for the discovery of new drugs?

Reductionism is alive and well in drug-discovery research. In that tradition, we continually improve experimental and computational methods for studying smaller and smaller aspects of biological systems. Although significant improvements continue to be made, are our efforts too narrowly focused? Suppose all error could be removed from these methods, would we then understand biological systems sufficiently well to design effective drugs? Currently, almost all drug research focuses on single targets. Should the process be expanded to include multiple targets? Recent efforts in this direction have lead to the emerging field of polypharmacology. This appears to be a move in the right direction, but how much polypharmacology is enough? As the complexity of the processes underlying polypharmacology increase will we be able to understand them and their inter-relationships? Is "new" mathematics unfamiliar in much of physics and chemistry research needed to accomplish this task? A number of these questions will be addressed in this paper, which focuses on issues and questions not answers to the drug-discovery conundrum.

Prospective Feasibility Trial for Genomics-Informed Treatment in Recurrent and Progressive Glioblastoma.

Purpose: Glioblastoma is an aggressive and molecularly heterogeneous cancer with few effective treatment options. We hypothesized that next-generation sequencing can be used to guide treatment recommendations within a clinically acceptable time frame following surgery for patients with recurrent glioblastoma.Experimental Design: We conducted a prospective genomics-informed feasibility trial in adults with recurrent and progressive glioblastoma. Following surgical resection, genome-wide tumor/normal exome sequencing and tumor RNA sequencing were performed to identify molecular targets for potential matched therapy. A multidisciplinary molecular tumor board issued treatment recommendations based on the genomic results, blood-brain barrier penetration of the indicated therapies, drug-drug interactions, and drug safety profiles. Feasibility of generating genomics-informed treatment recommendations within 35 days of surgery was assessed.Results: Of the 20 patients enrolled in the study, 16 patients had sufficient tumor tissue for analysis. Exome sequencing was completed for all patients, and RNA sequencing was completed for 14 patients. Treatment recommendations were provided within the study's feasibility time frame for 15 of 16 (94%) patients. Seven patients received treatment based on the tumor board recommendations. Two patients reached 12-month progression-free survival, both adhering to treatments based on the molecular profiling results. One patient remained on treatment and progression free 21 months after surgery, 3 times longer than the patient's previous time to progression. Analysis of matched nonenhancing tissue from 12 patients revealed overlapping as well as novel putatively actionable genomic alterations.Conclusions: Use of genome-wide molecular profiling is feasible and can be informative for guiding real-time, central nervous system-penetrant, genomics-informed treatment recommendations for patients with recurrent glioblastoma. Clin Cancer Res; 24(2); 295-305. ©2017 AACRSee related commentary by Wick and Kessler, p. 256.

Large compound databases for structure-activity relationships studies in drug discovery.

Large libraries of chemical compounds reflect the exponentially growing data-enrichment in drug discovery that trends towards fully automated informatics solutions to study structure - activity relationships by screening docked ligand candidates to biological target structures. We review otherwise disseminated user descriptions of mainly public databases with free access and also our integrated data mining tool GPDBnet for phyto-pharmacology.

Hit-directed nearest-neighbor searching.

This work describes a practical strategy used at Pharmacia for identifying compounds for follow-up screening following an initial HTS campaign against targets where no 3-D structural information is available and preliminary SAR models do not exist. The approach explicitly takes into account different representations of chemistry space and identifies compounds for follow-up screening that are likely to provide the best overall coverage of the chemistry spaces considered. Specifically, the work employs hit-directed nearest-neighbor (HDNN) searching of compound databases based upon a set of "probe compounds" obtained as hits in the preliminary high-throughput screens. Four different molecular representations that generate nearly unique chemistry spaces are used. The representations include 3-D, 2-D, 2-D topological BCUTs (2-DT) and molecular fingerprints derived from substructural fragments. In the case of the BCUT representations the NN searching is distance based, while in the case of molecular fingerprints a similarity-based measure is used. Generally, the results obtained differ significantly among all four methods, that is, the sets of NN compounds have surprisingly little overlap. Moreover, in all of the four chemistry space representations, a minimum of 3- to 4-fold enrichment in actives over random screening is observed even though the actives identified in each of the sets of NNs are in large measure unique. These results suggest that use of multiple searches based upon a variety of molecular representations provides an effective way of identifying more hits in HDNN searches of chemistry spaces than can be realized with single searches.

Evaluating the high-throughput screening computations.

The judges evaluated the submissions for the McMaster University High-Throughput Data-Mining and Docking Competition based on 3 criteria: identification of active compounds, percent enrichment, and overview of the competition. Using these metrics, 4 of the participating groups found meaningful enrichment, and 3 groups made perceptive comments about the general nature of the competition.

Assessment of the consistency of medicinal chemists in reviewing sets of compounds.

Medicinal chemists are frequently asked to review lists of compounds to assess their drug- or leadlike nature and to evaluate the suitability of lead compounds based on their "attractiveness" and/or synthetic feasibility as a basis for launching a drug-discovery campaign. It is often felt that one medicinal chemist's opinion is as good as any other, but is it? In an attempt to answer this question, an experiment was performed in conjunction with a recent compound acquisition program (CAP) conducted at Pharmacia. Historically, the CAP included a review of many thousands of compounds by medicinal chemists who eliminate anything deemed undesirable for any reason. In a review conducted in 2002, about 22 000 compounds requiring review by medicinal chemists were broken down into 11 lists of approximately 2000 compounds each. Unknown to the medicinal chemists, a subset of 250 compounds, previously rejected by a very experienced senior medicinal chemist, was added to each of the lists. Most of the 13 medicinal chemists who participated in this process reviewed two lists, although some only reviewed a single list and one reviewed three lists. Those compounds that were deemed unacceptable were recorded and tabulated in various ways to assess the consistency of the reviews. It was found that medicinal chemists were not very consistent in the compounds they rejected as being undesirable. The inconsistency arises from the subjective analysis that all humans utilize when considering "data sets" of any kind. This has important implications for pharmaceutical project teams where individual medicinal chemists review lists of primary screening hits to identify those compounds suitable for follow-up. Once a compound is removed from a list, it and other structurally similar compounds are effectively removed from further consideration. This can also have an impact on computational chemists who are developing models for assessing the desirability or attractiveness of different classes of compounds for lead discovery.

Molecular similarity measures.

Molecular similarity is a pervasive concept in chemistry. It is essential to many aspects of chemical reasoning and analysis and is perhaps the fundamental assumption underlying medicinal chemistry. Dissimilarity, the complement of similarity, also plays a major role in a growing number of applications of molecular diversity in combinatorial chemistry, high-throughput screening, and related fields. How molecular information is represented, called the representation problem, is important to the type of molecular similarity analysis (MSA) that can be carried out in any given situation. In this work, four types of mathematical structure are used to represent molecular information: sets, graphs, vectors, and functions. Molecular similarity is a pairwise relationship that induces structure into sets of molecules, giving rise to the concept of a chemistry space. Although all three concepts molecular similarity, molecular representation, and chemistry space are treated in this chapter, the emphasis is on molecular similarity measures. Similarity measures, also called similarity coefficients or indices, are functions that map pairs of compatible molecular representations, that is, representations of the same mathematical form, into real numbers usually, but not always, lying on the unit interval. This chapter presents a somewhat pedagogical discussion of many types of molecular similarity measures, their strengths and limitations, and their relationship to one another.