Maximum Common Substructure Research Articles

Knowledge-Based Strategy to Improve Ligand Pose Prediction Accuracy for Lead Optimization.

Accurately predicting how a small molecule binds to its target protein is an essential requirement for structure-based drug design (SBDD) efforts. In structurally enabled medicinal chemistry programs, binding pose prediction is often applied to ligands after a related compound's crystal structure bound to the target protein has been solved. In this article, we present an automated pose prediction protocol that makes extensive use of existing X-ray ligand information. It uses spatial restraints during docking based on maximum common substructure (MCS) overlap between candidate molecule and existing X-ray coordinates of the related compound. For a validation data set of 8784 docking runs, our protocol's pose prediction accuracy (80-82%) is almost two times higher than that of one unbiased docking method software (43%). To demonstrate the utility of this protocol in a project setting, we show its application in a chronological manner for a number of internal drug discovery efforts. The accuracy and applicability of this algorithm (>70% of cases) to medicinal chemistry efforts make this the approach of choice for pose prediction in lead optimization programs.

Efficient heuristics for maximum common substructure search.

Maximum common substructure search is a computationally hard optimization problem with diverse applications in the field of cheminformatics, including similarity search, lead optimization, molecule alignment, and clustering. Most of these applications have strict constraints on running time, so heuristic methods are often preferred. However, the development of an algorithm that is both fast enough and accurate enough for most practical purposes is still a challenge. Moreover, in some applications, the quality of a common substructure depends not only on its size but also on various topological features of the one-to-one atom correspondence it defines. Two state-of-the-art heuristic algorithms for finding maximum common substructures have been implemented at ChemAxon Ltd., and effective heuristics have been developed to improve both their efficiency and the relevance of the atom mappings they provide. The implementations have been thoroughly evaluated and compared with existing solutions (KCOMBU and Indigo). The heuristics have been found to greatly improve the performance and applicability of the algorithms. The purpose of this paper is to introduce the applied methods and present the experimental results.

Maximum common substructure-based data fusion in similarity searching.

Data fusion has been shown to work very well when applied to fingerprint-based similarity searching, yet little is known of its application to maximum common substructure (MCS)-based similarity searching. Two similarity search applications of the MCS will be focused on here. Typically, the number of bonds in the MCS, as well as the bonds in the two molecules being compared, are used in a similarity coefficient. The power of this technique can be extended using data fusion, where the MCS similarities of a set of reference molecules against one database molecule are fused. This "group fusion" technique forms the first application of the MCS in this work. The other application is that of the chemical hyperstructure. The hyperstructure concept is an alternative form of data fusion, being a hypothetical molecule that is constructed from the overlap of a set of existing molecules. This paper compares fingerprint group fusion (extended-connectivity fingerprints), MCS similarity group fusion, and hyperstructure similarity searching, and describes their relative merits and complementarity in virtual screening. It is concluded that the hyperstructure approach as implemented here is less generally effective than conventional fingerprint approaches.

CLCA: maximum common molecular substructure queries within the MetRxn database.

The challenge of automatically identifying the preserved molecular moieties in a chemical reaction is referred to as the atom mapping problem. Reaction atom maps provide the ability to locate the fate of individual atoms across an entire metabolic network. Atom maps are used to track atoms in isotope labeling experiments for metabolic flux elucidation, trace novel biosynthetic routes to a target compound, and contrast entire pathways for structural homology. However, rapid computation of the reaction atom mappings remains elusive despite significant research. We present a novel substructure search algorithm, canonical labeling for clique approximation (CLCA), with polynomial run-time complexity to quickly generate atom maps for all the reactions present in MetRxn. CLCA uses number theory (i.e., prime factorization) to generate canonical labels or unique IDs and identify a bijection between the vertices (atoms) of two distinct molecular graphs. CLCA utilizes molecular graphs generated by combining atomistic information on reactions and metabolites from 112 metabolic models and 8 metabolic databases. CLCA offers improvements in run time, accuracy, and memory utilization over existing heuristic and combinatorial maximum common substructure (MCS) search algorithms. We provide detailed examples on the various advantages as well as failure modes of CLCA over existing algorithms.

3D-QSAR modeling and molecular docking study on Mer kinase inhibitors of pyridine-substituted pyrimidines.

Mer kinase is a novel therapeutic target for many cancers, and overexpression of Mer receptor tyrosine kinase has been observed in several kinds of tumors. To deeply understand the structure-activity correlation of a series of pyridine/pyrimidine analogs as potent Mer inhibitors, a combined molecular docking and three-dimensional quantitative structure-activity relationship modeling was carried out. A comparative molecular similarity indices analysis model was developed based on the maximum common substructure alignment. The optimum model exhibited statistically significant results: the cross-validated correlation coefficient q2 was 0.599, and non-cross-validated r2 value was 0.984. Furthermore, the results of internal validation such as bootstrapping, Y-randomization as well as external validation (the external predictive correlation coefficient r2 ext = 0.728) confirmed the rationality and good predictive ability of the model. Using the crystal structure of Mer kinase, the selected pyridine/pyrimidine compounds were docked into the enzyme active site. Some key amino acid residues were determined, and hydrogen bonding and hydrophobic interactions between Mer kinase and inhibitors were identified. The satisfactory results from this study may aid in the research and development of novel potent Mer kinase inhibitors.

3D flexible alignment using 2D maximum common substructure: dependence of prediction accuracy on target-reference chemical similarity.

A protein-bound conformation of a target molecule can be predicted by aligning the target molecule on the reference molecule obtained from the 3D structure of the compound-protein complex. This strategy is called "similarity-based docking". For this purpose, we develop the flexible alignment program fkcombu, which aligns the target molecule based on atomic correspondences with the reference molecule. The correspondences are obtained by the maximum common substructure (MCS) of 2D chemical structures, using our program kcombu. The prediction performance was evaluated using many target-reference pairs of superimposed ligand 3D structures on the same protein in the PDB, with different ranges of chemical similarity. The details of atomic correspondence largely affected the prediction success. We found that topologically constrained disconnected MCS (TD-MCS) with the simple element-based atomic classification provides the best prediction. The crashing potential energy with the receptor protein improved the performance. We also found that the RMSD between the predicted and correct target conformations significantly correlates with the chemical similarities between target-reference molecules. Generally speaking, if the reference and target compounds have more than 70% chemical similarity, then the average RMSD of 3D conformations is <2.0 Å. We compared the performance with a rigid-body molecular alignment program based on volume-overlap scores (ShaEP). Our MCS-based flexible alignment program performed better than the rigid-body alignment program, especially when the target and reference molecules were sufficiently similar.

InSARa: intuitive and interactive SAR interpretation by reduced graphs and hierarchical MCS-based network navigation.

The analysis of Structure-Activity-Relationships (SAR) of small molecules is a fundamental task in drug discovery. Although a large number of methods are already published, there is still a strong need for novel intuitive approaches. The inSARa (intuitive networks for Structure-Activity Relationships analysis) method introduced herein takes advantage of the synergistic combination of reduced graphs (RG) and the intuitive maximum common substructure (MCS) concept. The main feature of the inSARa concept is a hierarchical network structure of clearly defined substructure relationships based on common pharmacophoric features. Thus, straightforward SAR interpretation is possible by interactive network navigation. When focusing on a set of active molecules at one single target, the resulting inSARa networks are shown to be valuable for various essential tasks in SAR analysis, such as the identification of activity cliffs or "activity switches", bioisosteric exchanges, common pharmacophoric features, or "SAR hotspots".

Development of pharmacophore similarity-based quantitative activity hypothesis and its applicability domain: applied on a diverse data-set of HIV-1 integrase inhibitors

Quantitative pharmacophore hypothesis combines the 3D spatial arrangement of pharmacophore features with biological activities of the ligand data-set and predicts the activities of geometrically and/or pharmacophoric similar ligands. Most pharmacophore discovery programs face difficulties in conformational flexibility, molecular alignment, pharmacophore features sampling, and feature selection to score models if the data-set constitutes diverse ligands. Towards this focus, we describe a ligand-based computational procedure to introduce flexibility in aligning the small molecules and generating a pharmacophore hypothesis without geometrical constraints to define pharmacophore space, enriched with chemical features necessary to elucidate common pharmacophore hypotheses (CPHs). Maximal common substructure (MCS)-based alignment method was adopted to guide the alignment of carbon molecules, deciphered the MCS atom connectivity to cluster molecules in bins and subsequently, calculated the pharmacophore similarity matrix with the bin-specific reference molecules. After alignment, the carbon molecules were enriched with original atoms in their respective positions and conventional pharmacophore features were perceived. Distance-based pharmacophoric descriptors were enumerated by computing the interdistance between perceived features and MCS-aligned ‘centroid’ position. The descriptor set and biological activities were used to develop support vector machine models to predict the activities of the external test set. Finally, fitness score was estimated based on pharmacophore similarity with its bin-specific reference molecules to recognize the best and poor alignments and, also with each reference molecule to predict outliers of the quantitative hypothesis model. We applied this procedure to a diverse data-set of 40 HIV-1 integrase inhibitors and discussed its effectiveness with the reported CPH model.

Introducing fuzziness into maximum common substructures for meaningful cluster characterisation

Arranging similar structures in clusters is one of the typical tasks of modern Chemoinformatics with high impact in HTS follow-up, generation of structure activity relationships (SAR) and selection of starting points for compound optimisation. Methods for cluster generation are as diverse as the structures which they are applied to [1], may they be e.g. similarity- or substructure-based. Typically, medicinal chemists tend to orientate themselves in structure subsets like clusters with the help of substructures, so-called scaffolds, which intuitively characterise the structural relationships between the molecules of the subset. In the case of substructure-based clustering, well established methods are existing for the generation of Maximum Common Substructures (MCS) which are present in all members of the structure population or a defined proportion thereof [2]. But in the case of similarity-based clusters, such MCS may either not be existing for the required dataset proportion or the common substructure may be so small that it is no longer representative and therefore meaningless. The approach presented here allows the generation of MCS also for similarity-based clusters with a given inherent structural diversity. It does so by generating an MCS of reduced graphs in a first step, followed by mapping atom and bond indexes of this reduced MCS onto the full structures and aggregation of atom and bond information for each indexed atom/bond. In a final step, query features of the MDL SDF format (atom lists, query bonds) are utilized to map aggregated element and bond information onto the reduced MCS. As a result, fuzziness in atom and bond information is added to the MCS which, although still being fully database-searchable, is more meaningful for the characterisation of clusters as it can cover larger parts of the full structures than a conventional MCS could do. The approach was implemented in Pipeline Pilot™ for proof of concept but is general enough to be transferred to other technical platforms as well.

Making the most of approximate maximum common substructure search

The maximum common substructure (MCS) problem is of great importance in multiple aspects of chemoinformatics. It has diverse applications ranging from lead prediction to automated reaction mapping and visual alignment of similar compounds. Many different algorithms have been developed [1], both exact and approximate. Since the MCS problem is NP-complete, the strict time constraints of most applications can only be realistically satisfied by fast and robust approximation methods. We developed two efficient heuristic algorithms. One is based on the popular approach of reducing the MCS problem to finding the maximum clique in the modular product of the two molecule graphs. The other is based on a new algorithm by Kawabata, called the build-up method [2]. We also incorporated other techniques, for example, the topological fingerprinting primarily used in substructure and similarity searching. We optimized our implementations for use in multiple applications developed at ChemAxon. In some applications, for example, hierarchical MCS-based clustering or similarity search in large databases, the algorithms are required to give close to optimal results in limited time. To meet these conflicting demands, our implementations were enhanced with strong heuristics. Upper bound calculation methods were also applied for screening and early termination purposes. In other applications, for example, reaction mapping or visual alignment, the challenge is that topological features must also be taken into account. Apart from the size of the found common substructure, the determined one-to-one correspondence between the atoms of the molecules is also very important. Effective heuristics were developed to guide the algorithms to prefer those solutions in which the relative positions of the common fragments of the input molecules are as similar as possible. Our implementations have been thoroughly tested and benchmarked. They have also been compared to publicly available solution methods, and integrated into different products at ChemAxon. This has shown that the presented MCS algorithms can adequately cover the conflicting requirements of typical applications. We present our methods and heuristics along with their effects on running time, memory usage, as well as the size and features of the result.

InSARa: intuitive single-target (large-scale) SAR interpretation and multi-target cross-reactivity analysis

inSARa (intuitive networks for Structure-Activity-Relationships analysis) was primarily developed with the objective to support the medicinal chemist in tackling SAR analysis and visualization of large data sets in a more intuitive way than fingerprint-based approaches [1]. The method takes advantage of the synergic combination of the reduced graph (RG) and the maximum common substructure (MCS) concept [2]. The main feature of the inSARa concept is a hierarchical network structure of clearly-defined substructure relationships based on common pharmacophoric features. Thus, straightforward SAR interpretation is possible by interactive network navigation. When focusing on a set of active molecules at one single target, the resulting inSARa networks were shown to be valuable for various essential tasks in SAR analysis, such as the identification of activity cliffs or activity switches, bioisosteric replacements or SAR hotspots. Based on the identification of nearest neighbours in the networks, the prediction of bioactivities is also possible. Furthermore, inSARa can be used to investigate similarities between different targets. Targets are compared based on the overlap of common pharmacophoric pattern (RG MCSs) of the corresponding inSARa networks. According to the similar property principle, similar ligands are expected to bind to similar targets [3]. Therefore, this ligand-based analysis not only revealed meaningful similarity relationships between the analysed targets but is also beneficial for the detection of potential off-target relationships and cross-reactivities. Especially when investigating targets where no structural information is available but a set of active ligands is known (e.g. GPCRs), this complementary approach can provide important knowledge for drug design.

Pharmacophore elucidation and 3D-QSAR analysis of a new class of highly potent inhibitors of acid ceramidase based on maximum common substructure and field fit alignment methods

Comparative molecular field analysis (CoMFA) and comparative similarity indices analysis (CoMSIA) studies have been carried on a series of 2,4-dioxopyrimidine-1-carboxamides as acid ceramidase inhibitors. Two alignment rules for the compounds were defined using maximum common substructure and field fit. The best orientation was then searched by all-orientation search strategy, to minimize the effect of the initial orientation of the structures. The Kennard Stone algorithm was used to divide the entire set into training (25 compounds) and test (7 compounds) sets. Pharmacophore model identification was also performed using DISCOtech algorithm and refinement was carried out using GASP, to highlight important structural features that could be responsible for the inhibitory activity. All constructed models showed appropriate statistical parameters in terms of q2 and rpred2. Based upon the information obtained from CoMFA, CoMSIA, and developed pharmacophore pattern, some key features that may be used to design new inhibitors for acid ceramidase have been identified.

Accuracy Assessment and Automation of Free Energy Calculations for Drug Design

As the free energy of binding of a ligand to its target is one of the crucial optimization parameters in drug design, its accurate prediction is highly desirable. In the present study we have assessed the average accuracy of free energy calculations for a total of 92 ligands binding to five different targets. To make this study and future larger scale applications possible we automated the setup procedure. Starting from user defined binding modes, the procedure decides which ligands to connect via a perturbation based on maximum common substructure criteria and produces all necessary parameter files for free energy calculations in AMBER 11. For the systems investigated, errors due to insufficient sampling were found to be substantial in some cases whereas differences in estimators (thermodynamic integration (TI) versus multistate Bennett acceptance ratio (MBAR)) were found to be negligible. Analytical uncertainty estimates calculated from a single free energy calculation were found to be much smaller than the sample standard deviation obtained from two independent free energy calculations. Agreement with experiment was found to be system dependent ranging from excellent to mediocre (RMSE = [0.9, 8.2, 4.7, 5.7, 8.7] kJ/mol). When restricting analyses to free energy calculations with sample standard deviations below 1 kJ/mol agreement with experiment improved (RMSE = [0.8, 6.9, 1.8, 3.9, 5.6] kJ/mol).

Algorithm for Reaction Classification

Reaction classification has important applications, and many approaches to classification have been applied. Our own algorithm tests all maximum common substructures (MCS) between all reactant and product molecules in order to find an atom mapping containing the minimum chemical distance (MCD). Recent publications have concluded that new MCS algorithms need to be compared with existing methods in a reproducible environment, preferably on a generalized test set, yet the number of test sets available is small, and they are not truly representative of the range of reactions that occur in real reaction databases. We have designed a challenging test set of reactions and are making it publicly available and usable with InfoChem's software or other classification algorithms. We supply a representative set of example reactions, grouped into different levels of difficulty, from a large number of reaction databases that chemists actually encounter in practice, in order to demonstrate the basic requirements for a mapping algorithm to detect the reaction centers in a consistent way. We invite the scientific community to contribute to the future extension and improvement of this data set, to achieve the goal of a common standard.

Lead optimization mapper: automating free energy calculations for lead optimization

Alchemical free energy calculations hold increasing promise as an aid to drug discovery efforts. However, applications of these techniques in discovery projects have been relatively few, partly because of the difficulty of planning and setting up calculations. Here, we introduce lead optimization mapper, LOMAP, an automated algorithm to plan efficient relative free energy calculations between potential ligands within a substantial library of perhaps hundreds of compounds. In this approach, ligands are first grouped by structural similarity primarily based on the size of a (loosely defined) maximal common substructure, and then calculations are planned within and between sets of structurally related compounds. An emphasis is placed on ensuring that relative free energies can be obtained between any pair of compounds without combining the results of too many different relative free energy calculations (to avoid accumulation of error) and by providing some redundancy to allow for the possibility of error and consistency checking and provide some insight into when results can be expected to be unreliable. The algorithm is discussed in detail and a Python implementation, based on both Schrödinger's and OpenEye's APIs, has been made available freely under the BSD license.

FmcsR: mismatch tolerant maximum common substructure searching in R

The ability to accurately measure structural similarities among small molecules is important for many analysis routines in drug discovery and chemical genomics. Algorithms used for this purpose include fragment-based fingerprint and graph-based maximum common substructure (MCS) methods. MCS approaches provide one of the most accurate similarity measures. However, their rigid matching policies limit them to the identification of perfect MCSs. To eliminate this restriction, we introduce a new mismatch tolerant search method for identifying flexible MCSs (FMCSs) containing a user-definable number of atom and/or bond mismatches. The fmcsR package provides an R interface, with the time-consuming steps of the FMCS algorithm implemented in C++. It includes utilities for pairwise compound comparisons, structure similarity searching, clustering and visualization of MCSs. In comparison with an existing MCS tool, fmcsR shows better time performance over a wide range of compound sizes. When mismatching of atoms or bonds is turned on, the compute times increase as expected, and the resulting FMCSs are often substantially larger than their strict MCS counterparts. Based on extensive virtual screening (VS) tests, the flexible matching feature enhances the enrichment of active structures at the top of MCS-based similarity search results. With respect to overall and early enrichment performance, FMCS outperforms most of the seven other VS methods considered in these tests. fmcsR is freely available for all common operating systems from the Bioconductor site (http://www.bioconductor.org/packages/devel/bioc/html/fmcsR.html). thomas.girke@ucr.edu. Supplementary data are available at Bioinformatics online.

Alzheimer’s Disease: Identification and Development of β‐Secretase (BACE‐1) Binding Fragments and Inhibitors by Dynamic Ligation Screening (DLS)

The application of dynamic ligation screening (DLS), a methodology for fragment-based drug discovery (FBDD), to the aspartic protease β-secretase (BACE-1) is reported. For this purpose, three new fluorescence resonance energy transfer (FRET) substrates were designed and synthesized. Their kinetic parameters (Vmax , KM , and kcat ) were determined and compared with a commercial substrate. Secondly, a peptide aldehyde was designed as a chemically reactive inhibitor (CRI) based on the Swedish mutation substrate sequence. Incubation of this CRI with the protease, a FRET substrate, and one amine per well taken from an amine library, which was assembled by a maximum common substructure (MCS) approach, revealed the fragment 3-(3-aminophenyl)-2H-chromen-2-one (1) to be a competitive BACE-1 inhibitor that enhanced the activity of the CRI. Irreversibly formed fragment combination products of 1 with the initial peptide sequence were active and confirmed the targeting of the active site through the ethane-1,2-diamine isostere. Finally, structure-assisted combination of fragment 1 with secondary fragments that target the S1 site in hit optimization yielded novel, entirely fragment-based BACE-1 inhibitors with up to 30-fold improved binding affinity. Interactions with the protein were explained by molecular modeling studies, which indicate that the new fragment combinations interact with the catalytic aspartic acid dyad, as well as with the adjacent binding sites required for potency.

Automatic reaction mapping and reaction center detection

AbstractA reaction center is the part of a chemical reaction that undergoes changes, the heart of the chemical reaction. The reaction atom–atom mapping indicates which reactant atom becomes which product atom during the reaction. Automatic reaction mapping and reaction center detection are of great importance in many applications, such as developing chemical and biochemical reaction databases and studying reaction mechanisms. Traditional reaction mapping algorithms are either based on extended‐connectivity or maximum common substructure (MCS) algorithms. With the development of several biochemical reaction databases (such as KEGG database) and increasing interest in studying metabolic pathways in recent years, several novel reaction mapping algorithms have been developed to serve the new needs. Most of the new algorithms are optimization based, designed to find optimal mappings with the minimum number of broken and formed bonds. Some algorithms also incorporate the chemical knowledge into the searching process in the form of bond weights. Some new algorithms showed better accuracy and performance than the MCS‐based method.WIREs Comput Mol Sci2013, 3:560–593. doi: 10.1002/wcms.1140This article is categorized under:Computer and Information Science &gt; Chemoinformatics

Combination of fingerprints and MCS-based (inSARa) networks for Structure-Activity-Relationship analysis

Structure-Activity-Relationship (SAR) analysis of small molecules is a fundamental and challenging task in drug discovery. The knowledge of these relationships between chemical structure and bioactivity is of high value for the medicinal chemist, e.g., in the lead-optimization process or de-novo-design. In order to analyse SARs, the recognition of molecular similarities is a crucial step due to the assumption that molecules with similar properties or features exhibit similar biological activities [1]. Fingerprint-based similarity is an established concept which is also the basis for the Network-like Similarity Graph approach (NSG) [2]. Advantages of fingerprints are that they are fast to compute and well suited for processing large data sets. On the other hand, the interpretation of the results is difficult and the concept appears less intuitive to a medicinal chemist. A promising and more straightforward alternative is the maximum common substructure (MCS). One drawback is the high computational cost which prevents, for instance, the analysis of large HTS data sets. To tackle these limitations our new method, inSARa (intuitive networks for Structure-Activity-Relationships analysis), combines MCS similarity with the reduced graph (RG) concept [3] for SAR network generation. By using RGs, a higher degree of abstraction is provided as they supply a conceptual representation of pharmacophoric features of a molecule by reducing groups of atoms to single pseudoatoms while retaining the original topology. Moreover, the probability of scaffold hopping and identifying structurally less similar bioisosteric substructures is increased. Recently, we could show that fingerprints and inSARa provide complementary information for SAR analysis. By combining both approaches, synergetic effects are expected. This will be demonstrated in this contribution.

FMCS: a novel algorithm for the multiple MCS problem

Clustering and classification of large-scale chemical data are essential for navigation, analysis and knowledge discovery in a wide variety of chemical application domains. The maximum common structure (MCS) for a group of compounds is an important element of such classification, providing insight into activity patterns and enabling scaffold alignment for a more consistent 2D depiction. Most modern, exact MCS implementations use back-tracking [1] or clique detection [2], and handle the multiple MCS problem by recursive reduction to successive pairwise maximal common substructure searches [3]. We present fmcs, which implements a novel multiple MCS algorithm based on subgraph enumeration and subgraph isomorphism testing [4,5] and with algorithm improvements and heuristics which make it competitive to the standard methods. MCS performance evaluation is very sensitive to the test set, so we have developed several reference benchmarks based on ChEMBL-13, including randomly selected pairs of structures, and randomly selected structures with their k = 2, k = 10, and k = 100 nearest neighbors. We use these benchmarks to compare fmcs to SMSD [6] and Indigo’s scaffold detector [7]. Most differences are due to chemistry perception and timeout errors. The fmcs performance, written in Python using the RDKit C++ toolkit [8], is currently between 0.3x and 1.2x the performance of the Indigo implementation in C++. We also cross-validated the fmcs algorithm with the manually curated ChEBI structure ontology classification [9] and characterized the differences. We identified limitations with fmcs, such as with tautomer perception and structural classes that fmcs cannot handle, and problems with ChEBI, such as misclassifications and classifications that are not, structurally speaking, strictly hierarchical.