Racemisation in Chemistry and Biology.

The two enantiomers of a compound often have profoundly different biological properties and thus their liability to racemisation in aqueous solutions is an important piece of information. The authors reviewed the available data concerning the process of racemisation in vivo, in the presence of biological molecules (e.g., racemase enzymes, serum albumin, cofactors and derivatives) and under purely chemical but aqueous conditions (acid, base and other aqueous systems). Mechanistic studies are described critically in light of reported kinetic data. The types of experimental measurement that can be used to effectively determine rate constants of racemisation in various conditions are discussed and the data they provide is summarised. The proposed origins of enzymatic racemisation are presented and suggest ways to promote the process that are different from processes taking place in bulk water. Experimental and computational studies that provide understanding and quantitative predictions of racemisation risk are also presented.

Alignment-Free Molecular Shape Comparison Using Spectral Geometry: The Framework.

A framework is presented for the calculation of novel alignment-free descriptors of molecular shape. The methods are based on the technique of spectral geometry which has been developed in the field of computer vision where it has shown impressive performance for the comparison of deformable objects such as people and animals. Spectral geometry techniques encode shape by capturing the curvature of the surface of an object into a compact, information-rich representation that is alignment-free while also being invariant to isometric deformations, that is, changes that do not distort distances over the surface. Here, we adapt the technique to the new domain of molecular shape representation. We describe a series of parametrization steps aimed at optimizing the method for this new domain. Our focus here is on demonstrating that the basic approach is able to capture a molecular shape into a compact and information-rich descriptor. We demonstrate improved performance in virtual screening over a more established alignment-free method and impressive performance compared to a more accurate, but much more computationally demanding, alignment-based approach.

ReFlex3D: Refined Flexible Alignment of Molecules Using Shape and Electrostatics.

We present an algorithm, ReFlex3D, for the refinement of flexible molecular alignments based on their three-dimensional shape and electrostatic properties. The algorithm is designed to be used with fast conformer generators to refine an initial overlay between two molecules and thus to obtain improved overlaps as judged by an increase in calculated similarity values. ReFlex3D is open-source and built as a python package working in combination with the OEChem Toolkit. As such it can readily be implemented in existing workflows ranging from the selection of compounds from a virtual screening campaign to the construction of similarity based prediction models to estimate binding affinities. We evaluate ReFlex3D against the AstraZeneca Validation Test Set and illustrate its potential within a predictive model compared to an established method (Posit).

Quantitative Prediction of Rate Constants for Aqueous Racemization To Avoid Pointless Stereoselective Syntheses.

Racemization has a large impact upon the biological properties of molecules but the chemical scope of compounds with known rate constants for racemization in aqueous conditions was hitherto limited. To address this remarkable blind spot, we have measured the kinetics for racemization of 28 compounds using circular dichroism and 1 H NMR spectroscopy. We show that rate constants for racemization (measured by ourselves and others) correlate well with deprotonation energies from quantum mechanical (QM) and group contribution calculations. Such calculations thus provide predictions of the second-order rate constants for general-base-catalyzed racemization that are usefully accurate. When applied to recent publications describing the stereoselective synthesis of compounds of purported biological value, the calculations reveal that racemization would be sufficiently fast to render these expensive syntheses pointless.

Investigation of the use of spectral clustering for the analysis of molecular data.

Spectral clustering involves placing objects into clusters based on the eigenvectors and eigenvalues of an associated matrix. The technique was first applied to molecular data by Brewer [J. Chem. Inf. Model. 2007, 47, 1727-1733] who demonstrated its use on a very small dataset of 125 COX-2 inhibitors. We have determined suitable parameters for spectral clustering using a wide variety of molecular descriptors and several datasets of a few thousand compounds and compared the results of clustering using a nonoverlapping version of Brewer's use of Sarker and Boyer's algorithm with that of Ward's and k-means clustering. We then replaced the exact eigendecomposition method with two different approximate methods and concluded that Singular Value Decomposition is the most appropriate method for clustering larger compound collections of up to 100,000 compounds. We have also used spectral clustering with the Tversky coefficient to generate two sets of clusters linked by a common set of eigenvalues and have used this novel approach to cluster sets of fragments such as those used in fragment-based drug design.

An extensive and diverse set of molecular overlays for the validation of pharmacophore programs.

The pharmacophore hypothesis plays a central role in both the design and optimization of drug-like ligands. Pharmacophore patterns are invoked to explain the binding affinity of ligands and to enable the design of chemically distinct scaffolds that show affinity for a protein target of interest. The importance of pharmacophores in rationalizing ligand affinity has led to numerous algorithms that seek to overlay ligands based on their pharmacophoric features. All such algorithms must be validated with respect to known ligand overlays, usually by extracting ligand overlay sets from the Protein Data Bank (PDB). This validation step creates the problem of which of the known overlays to select and from which proteins. The large number of structures and protein families in the PDB makes it difficult to establish a definitive overlay set; as a result, validation studies have rarely employed the same data sets. We have therefore undertaken an exhaustive analysis of the RCSB PDB to identify 121 distinct ligand overlay sets. We have defined a robust protein overlay protocol, which is free from subjective interpretation over which residues to include, and we have analyzed each overlay set on the basis of whether they provide evidence for the pharmacophore hypothesis. Our final data set spans a broad range of structural types and degrees of difficulty and includes overlays that any algorithm should be able to reproduce, as well as some for which there is very weak evidence for a conserved pharmacophore at all. We provide this set in the hope that it will prove definitive, at least until the PDB is greatly enriched with further structures or with radically different protein folds and families. Upon publication, the data set will be available for free download from the Web site of the Cambridge Crystallographic Data Centre.

A system for encoding and searching Markush structures.

The encoding and searching of generic chemical structures, so-called Markush structures, have received little attention in the literature of late. The ability to encode and search these complex entities is of use in various branches of chemoinformatics. We describe a general language for encoding Markush structures and algorithms for searching them and give three examples of the utility of such a system: development of general Free-Wilson analyses of chemical series, detection of controlled substances within a large database of molecular structures, and searching of large databases of virtual compounds.

Development and validation of an improved algorithm for overlaying flexible molecules.

A program for overlaying multiple flexible molecules has been developed. Candidate overlays are generated by a novel fingerprint algorithm, scored on three objective functions (union volume, hydrogen-bond match, and hydrophobic match), and ranked by constrained Pareto ranking. A diverse subset of the best ranked solutions is chosen using an overlay-dissimilarity metric. If necessary, the solutions can be optimised. A multi-objective genetic algorithm can be used to find additional overlays with a given mapping of chemical features but different ligand conformations. The fingerprint algorithm may also be used to produce constrained overlays, in which user-specified chemical groups are forced to be superimposed. The program has been tested on several sets of ligands, for each of which the true overlay is known from protein-ligand crystal structures. Both objective and subjective success criteria indicate that good results are obtained on the majority of these sets.

Multiobjective optimization of pharmacophore hypotheses: bias toward low-energy conformations.

Two methods are described for biasing conformational search during pharmacophore elucidation using a multiobjective genetic algorithm (MOGA). The MOGA explores conformation on-the-fly while simultaneously aligning a set of molecules such that their pharmacophoric features are maximally overlaid. By using a clique detection method to generate overlays of precomputed conformations to initialize the population (rather than starting from random), the speed of the algorithm has been increased by 2 orders of magnitude. This increase in speed has enabled the program to be applied to greater numbers of molecules than was previously possible. Furthermore, it was found that biasing the conformations explored during search time to those found in the Cambridge Structural Database could also improve the quality of the results.

Design of compound libraries for fragment screening.

Approaches to the design of libraries for fragment screening are illustrated with reference to a 20 k generic fragment screening library and a 1.2 k generic NMR screening library. Tools and methods for library design that have been developed within AstraZeneca are described, including Foyfi fingerprints and the Flush program for neighborhood characterization. It will be shown how Flush and the BigPicker, which selects maximally diverse sets of compounds, are used to apply the Core and Layer method for library design. Approaches to partitioning libraries into cocktails are also described.

The problem of racemization in drug discovery and tools to predict it.

INTRODUCTION: Racemization has long been an ignored risk in drug development, probably because of a lack of convenient access to good tools for its detection and an absence of methods to predict racemization risk. As a result, the potential effects of racemization have been systematically underestimated. Areas covered: Herein, the potential effects of racemization are discussed through a review of drugs for which activity and side effects for both enantiomers are known. Subsequently, drugs known to racemize are discussed and the authors review methods to predict racemization risk. Application of a method quantitatively predicting racemization risk to databases of compounds from the medicinal chemistry literature shows that success in clinical trials is negatively correlated with racemization risk. Expert opinion: It is envisioned that a quantitative method of predicting racemization risk will remove a blind spot from the drug development pipeline. Removal of the blind spot will make drug development more efficient and result in less late-stage attrition of the drug pipeline.

Bioisosteric Replacements Extracted from High-Quality Structures in the Protein Databank.

Bioisosterism is an important concept in the lead optimisation phase of drug discovery where the aim is to make modifications to parts of a molecule in order to improve some properties while maintaining others. We present an analysis of bioisosteric fragments extracted from the ligands in an established data set consisting of 121 protein targets. A pairwise analysis is carried out of all ligands for a given target. The ligands are fragmented using the BRICS fragmentation scheme and a pair of fragments is deemed to be bioisosteric if they occupy a similar volume of the protein binding site. We consider two levels of generality, one which does not consider the number of attachment points in the fragments and a more restricted case in which both fragments are required to have the same number of attachments. We investigate the extent to which the bioisosteric pairs that are found are common across different target.

Matched molecular pairs as a guide in the optimization of pharmaceutical properties; a study of aqueous solubility, plasma protein binding and oral exposure.

By identifying every pair of molecules that differ only by a particular, well-defined, structural transformation in a database of measured properties and computing the corresponding change in property, we obtain an overview of the effect that structural change has upon the property and set an expectation for what will happen when that transformation is applied elsewhere. The mean change indicates the expected magnitude of the change in the property and the number of cases in which the property increases give the probability that the structural transformation will cause the property to increase. Outliers indicate potential ways of avoiding the general trend. Comparing to changes in lipophilicity highlights structural transformations that have unusual effects, some of which can be explained by conformational changes. In this paper, we focus upon the effects on aqueous solubility, plasma protein binding and oral exposure of adding substituents to aromatic rings and methylating heteroatoms.

Identification of compounds with nanomolar binding affinity for checkpoint kinase-1 using knowledge-based virtual screening.

A virtual screen of a subsection of the AstraZeneca compound collection was performed for checkpoint kinase-1 (Chk-1 kinase) using a knowledge-based strategy. This involved initial filtering of the compound collection by application of generic physical properties followed by removal of compounds with undesirable chemical functionality. Subsequently, a 3-D pharmacophore screen for compounds with kinase binding motifs was applied. A database of approximately 200K compounds remained for docking into the active site of Chk-1 kinase, using the FlexX-Pharm program. For each compound that docked successfully into the binding site, up to 100 poses were saved. These poses were then postfiltered using a customized consensus scoring scheme for a kinase, followed by visual inspection of a selection of the docked compounds. This resulted in 103 compounds being ordered for testing in the project assay, and 36 of these (corresponding to four chemical classes) were found to inhibit the enzyme in a dose-response fashion with IC(50) values ranging from 110 nM to 68 microM.

Representing clusters using a maximum common edge substructure algorithm applied to reduced graphs and molecular graphs.

Chemical databases are routinely clustered, with the aim of grouping molecules which share similar structural features. Ideally, medicinal chemists are then able to browse a few representatives of the cluster in order to interpret the shared activity of the cluster members. However, when molecules are clustered using fingerprints, it may be difficult to decipher the structural commonalities which are present. Here, we seek to represent a cluster by means of a maximum common substructure based on the shared functionality of the cluster members. Previously, we have used reduced graphs, where each node corresponds to a generalized functional group, as topological molecular descriptors for virtual screening. In this work, we precluster a database using any clustering method. We then represent the molecules in a cluster as reduced graphs. By repeated application of a maximum common edge substructure (MCES) algorithm, we obtain one or more reduced graph cluster representatives. The sparsity of the reduced graphs means that the MCES calculations can be performed in real time. The reduced graph cluster representatives are readily interpretable in terms of functional activity and can be mapped directly back to the molecules to which they correspond, giving the chemist a rapid means of assessing potential activities contained within the cluster. Clusters of interest are then subject to a detailed R-group analysis using the same iterated MCES algorithm applied to the molecular graphs.

Scaffold hopping using clique detection applied to reduced graphs.

Similarity-based methods for virtual screening are widely used. However, conventional searching using 2D chemical fingerprints or 2D graphs may retrieve only compounds which are structurally very similar to the original target molecule. Of particular current interest then is scaffold hopping, that is, the ability to identify molecules that belong to different chemical series but which could form the same interactions with a receptor. Reduced graphs provide summary representations of chemical structures and, therefore, offer the potential to retrieve compounds that are similar in terms of their gross features rather than at the atom-bond level. Using only a fingerprint representation of such graphs, we have previously shown that actives retrieved were more diverse than those found using Daylight fingerprints. Maximum common substructures give an intuitively reasonable view of the similarity between two molecules. However, their calculation using graph-matching techniques is too time-consuming for use in practical similarity searching in larger data sets. In this work, we exploit the low cardinality of the reduced graph in graph-based similarity searching. We reinterpret the reduced graph as a fully connected graph using the bond-distance information of the original graph. We describe searches, using both the maximum common induced subgraph and maximum common edge subgraph formulations, on the fully connected reduced graphs and compare the results with those obtained using both conventional chemical and reduced graph fingerprints. We show that graph matching using fully connected reduced graphs is an effective retrieval method and that the actives retrieved are likely to be topologically different from those retrieved using conventional 2D methods.

Lead hopping using SVM and 3D pharmacophore fingerprints.

The combination of 3D pharmacophore fingerprints and the support vector machine classification algorithm has been used to generate robust models that are able to classify compounds as active or inactive in a number of G-protein-coupled receptor assays. The models have been tested against progressively more challenging validation sets where steps are taken to ensure that compounds in the validation set are chemically and structurally distinct from the training set. In the most challenging example, we simulate a lead-hopping experiment by excluding an entire class of compounds (defined by a core substructure) from the training set. The left-out active compounds comprised approximately 40% of the actives. The model trained on the remaining compounds is able to recall 75% of the actives from the "new" lead series while correctly classifying >99% of the 5000 inactives included in the validation set.