Aromatic interactions are a commonly encountered geometrical feature in crystal structures indicative of a strong adhesion between components. Aromatics Analyser 2 is a program that highlights many types of ring pairs in a crystal structure that have relative positioning consistent with a strong aromatic interaction. The program functions through the CCDC crystal structure visualiser, Mercury, for interactive analysis, and the CSD Python API for large data analysis.

New informatics-based molecular descriptors are introduced that provide numeric measures of the crystallographically accessible conformational space (determined by rotamers and rings) of small organic molecules. The measures are based on observed small organic molecule crystal structures in the Cambridge Structural Database (CSD). The continuous measures provide a quantitative assessment of the rotatability of covalent bonds in molecules in the context of crystal structures, which are traditionally classified in a discrete binary way, either as rotatable or as completely non-rotatable. Two ensemble combinations of such rotatability scores are introduced. The ensemble scores adjust for molecular topology and provide upper and lower bounds for the size of the configuration space accessible to a molecule in the context of a crystal structure. The accessible molecular configuration space is fundamental to understanding molecular flexibility.

We present the protolysis-targeting chimera (PROTAC) Conformer Generator, a fast and knowledge-based tool for generating robust conformational ensembles of PROTACs and other chimeric degraders. The modeling protocol integrates conformer generation, rigid-body ternary complex (TC) assembly, and conformational sampling strategies that address the inherent flexibility and complexity of these molecules. Each modeled TC is evaluated using a clash-score and a surface-score, designed to prioritize sterically and geometrically plausible models with favorable protein surface interactions. The protocol was validated using experimentally determined PROTAC-mediated TC structures from the Protein Data Bank and "PROTAC-like" structures from the Cambridge Structural Database, demonstrating accuracy across diverse systems. The results show that the PROTAC Conformer Generator can reliably reproduce experimental conformations and generate simple TC models that recapitulate the relative orientations between E3 ubiquitin ligase and the protein of interest as observed in protein crystal structures. This robust validation supports the method's reliability and establishes a reference framework for degrader modeling studies. The PROTAC Conformer Generator provides a structured and validated workflow for modeling and assessing degrader conformations and ternary complexes, enabling rapid ensemble generation and downstream integration into relevant early stage drug design pipelines.

Since its first release in 2016, the Cambridge Structural Database Python application programming interface (CSD Python API) has seen steady uptake within the community that the Cambridge Crystallographic Data Centre serves. This article reviews the history of scripting interfaces, demonstrating the need, and then briefly outlines the technical structure of the API. It describes the reach of the CSD Python API, provides a selected review of its impact and gives some illustrative examples of what scientists can do with it. The article concludes with speculation as to how such endeavours will evolve over the next decade.

Three-dimensional (3D) deep molecular generative models offer the advantage of goal-directed generation based on 3D-dependent properties, such as binding affinity for structure-based design within binding pockets. Traditional benchmarks created to evaluate SMILES or molecular graphs generators, such as GuacaMol or MOSES, are limited to evaluate 3D generators as they do not assess the quality of the generated molecular conformation. In this work, we hence developed GenBench3D, which implements a new benchmark for models producing molecules within a binding pocket. Our main contribution is the Validity3D metric, evaluating the conformation quality using the likelihood of bond lengths and valence angles based on reference values observed in the Cambridge Structural Database. The LiGAN, 3D-SBDD, Pocket2Mol, TargetDiff, DiffSBDD and ResGen models were benchmarked. We show that only between 0 generated molecules have valid conformations. Performing local relaxation of generated molecules in the pocket considerably improved the Validity3D for all models by a minimum increase of 40 of valid relaxed molecules shows on average higher Vina score (i.e. worse) than the set of raw generated molecules, indicating that the binding affinity of raw generated molecules might be overestimated. Using the other scoring functions, that give higher importance to ligand strain, only yield improved scores when using valid relaxed molecules. Using valid relaxed molecules, TargetDiff and Pocket2Mol show better median Vina, Glide and Gold PLP scores than other models. We have publicly released GenBench3D on GitHub for broader use: https://github.com/bbaillif/genbench3d

Identifying bioactive conformations of small molecules is an essential process for virtual screening applications relying on three-dimensional structure such as molecular docking. For most small molecules, conformer generators retrieve at least one bioactive-like conformation, with an atomic root-mean-square deviation (ARMSD) lower than 1 Å, among the set of low-energy conformers generated. However, there is currently no general method to prioritise these likely target-bound conformations in the ensemble. In this work, we trained atomistic neural networks (AtNNs) on 3D information of generated conformers of a curated subset of PDBbind ligands to predict the ARMSD to their closest bioactive conformation, and evaluated the early enrichment of bioactive-like conformations when ranking conformers by AtNN prediction. AtNN ranking was compared with bioactivity-unaware baselines such as ascending Sage force field energy ranking, and a slower bioactivity-based baseline ranking by ascending Torsion Fingerprint Deviation to the Maximum Common Substructure to the most similar molecule in the training set (TFD2SimRefMCS). On test sets from random ligand splits of PDBbind, ranking conformers using ComENet, the AtNN encoding the most 3D information, leads to early enrichment of bioactive-like conformations with a median BEDROC of 0.29 ± 0.02, outperforming the best bioactivity-unaware Sage energy ranking baseline (median BEDROC of 0.18 ± 0.02), and performing on a par with the bioactivity-based TFD2SimRefMCS baseline (median BEDROC of 0.31 ± 0.02). The improved performance of the AtNN and TFD2SimRefMCS baseline is mostly observed on test set ligands that bind proteins similar to proteins observed in the training set. On a more challenging subset of flexible molecules, the bioactivity-unaware baselines showed median BEDROCs up to 0.02, while AtNNs and TFD2SimRefMCS showed median BEDROCs between 0.09 and 0.13. When performing rigid ligand re-docking of PDBbind ligands with GOLD using the 1% top-ranked conformers, ComENet ranked conformers showed a higher successful docking rate than bioactivity-unaware baselines, with a rate of 0.48 ± 0.02 compared to CSD probability baseline with a rate of 0.39 ± 0.02. Similarly, on a pharmacophore searching experiment, selecting the 20% top-ranked conformers ranked by ComENet showed higher hit rate compared to baselines. Hence, the approach presented here uses AtNNs successfully to focus conformer ensembles towards bioactive-like conformations, representing an opportunity to reduce computational expense in virtual screening applications on known targets that require input conformations.

In this work, libraries of crystallographically determined bond lengths, valence angles and torsion angles derived from the Cambridge Structural Database (CSD) are used to generate a score to validate an individual bond length, valence angle or torsion angle. A single score can also be derived for each ensemble of bond lengths, ensemble of valence angles and ensemble of torsion angles in the molecule. The data used to compute the score for the individual parts of the molecule are also used to compute a single value assessing the specific conformation relative to the geometric molecular data contained in the CSD. Such a single-number summary may be used, for example, to validate geometries of modelled molecules or molecular conformations in experimentally or computationally determined crystal structures. An example of how the score can be applied to an individual conformation is provided. Validation of the new score is performed by comparison with existing software for validating molecular geometries ( Mogul ) and a comparison with the CSD Conformer Generator scores is also carried out. Further application to the experimental standard deviation of C—C bonds of CSD entries is demonstrated, and the Platinum diverse data set is also studied.

Deep generative models have gained recent popularity for chemical design. Many of these models have historically operated in 2D space; however, more recently explicit 3D molecular generative models have become of interest, which are the topic of this article. Dozens of published models have been developed in the last few years to generate molecules directly in 3D, outputting both the atom types and coordinates, either in one-shot or adding atoms or fragments step-by-step. These 3D generative models can also be guided by structural information such as a binding pocket representation to successfully generate molecules with docking score ranges similar to known actives, but still showing lower computational efficiency and generation throughput than 1D/2D generative models and sometimes producing unrealistic conformations. We advocate for a unified benchmark of metrics to evaluate generation and propose perspectives to be addressed in next implementations.

The program Mercury , developed at the Cambridge Crystallographic Data Centre, was originally designed primarily as a crystal structure visualization tool. Over the years the fields and scientific communities of chemical crystallography and crystal engineering have developed to require more advanced structural analysis software. Mercury has evolved alongside these scientific communities and is now a powerful analysis, design and prediction platform which goes a lot further than simple structure visualization.

The value of a hydrogen bond network prediction model was improved using a tool to increase prediction trust. Its accuracy could be improved up to 73% or 89% with the compromise that only 34% and 8% of the test examples could be predicted.

The fundamental understanding of crystal structures and the structural drivers for stability is a key topic within chemical crystallography, as well being critical in an industrial context, particularly for the pharmaceutical and agrochemical industries.Here at the CCDC, we have been developing a suite of structure-based methods to help scientists better understand their crystal forms; to assess the stability and predict likely outcomes.Currently, the community's solid form analysis methods focus mostly on analysing the quantity and quality of hydrogen bonds.This has been shown as a reasonably reliable indicator of solid form stability in systems where hydrogen bonding is present and dominant.There are, however, many solid forms where hydrogen bonds do not dominate, or are missing entirely.In these cases, the ability to analyse additional types of interactions is needed to rapidly provide a more holistic assessment of a crystal structure.Aromatic interactions are known to be another important interaction type for solid form stability and particularly so when hydrogen bonds are weaker, or not present.We have developed a method to visualise and assess the strength of aromatic interactions in the context of a crystal structure, without needing to manually measure any geometric parameters.Using a sophisticated machinelearning approach, trained on high-level DFT-D3 energy calculations of benzene-benzene dimers, we can score phenyl ring-ring interactions quickly just from the observed geometry.The addition of effective assessment of phenyl-phenyl interactions to our solid form analysis toolbox takes us a big step closer towards more complete understanding of solid forms.

Fast generation of plausible molecular conformations is central to molecular modeling. This paper presents an approach to conformer generation that makes extensive use of the information available in the Cambridge Structural Database. By using geometric distributions derived from the Cambridge Structural Database, it is possible to create biologically relevant conformations in the majority of cases analyzed. The paper compares the performance of the approach with previously published evaluations, and presents some cases where the method fails. The method appears to show significantly improved performance in reproduction of the conformations of structures observed in the Cambridge Structural Database and the Protein Data Bank as compared to other published methods of a similar speed.

Certain broad low-energy peaks caused by a single partial wave in total cross sections are explained in terms of phase shifts. Such peaks have been associated with the real part of a Regge pole trajectory, having a maximum near an integer value of the angular momentum quantum number. At the peak energies, the pertinent partial-wave phase shift was shown to have a local maximum near a value π / 2 modulo π. This implies no time delay in the semiclassical context. The phenomenon is a quantum effect, lacking a semiclassical interpretation.

Successful methodologies for theoretical crystal structure prediction (CSP) on flexible pharmaceutical-like organic molecules explore the lattice energy surface to find a set of plausible crystal structures. The initial search stages of CSP studies use relatively simple lattice energy approximations as hundreds of thousands of minima have to be considered. These generated crystal structures often have poor molecular geometries, as well as inaccurate lattice energy rankings, and performing reasonably accurate but computationally affordable optimisations of the crystal structures generated in a search would be highly desirable. Here, we seek to explore whether semi-empirical quantum-mechanical methods can perform this task. We employed the dispersion-corrected tight-binding Hamiltonian (DFTB3-D3) to relax all the inter- and intra-molecular degrees of freedom of several thousands of generated crystal structures of five pharmaceutical-like molecules, saving a large amount of computational effort compared to earlier studies. The computational cost scales better with molecular size and flexibility than other CSP methods, suggesting that it could be extended to even larger and more flexible molecules. On average, this optimisation improved the average reproduction of the eight experimental crystal structures (RMSD15) and experimental conformers (RMSD1) by 4% and 23%, respectively. The intermolecular interactions were then further optimised using distributed multipoles, derived from the molecular wave-functions, to accurately describe the electrostatic components of the intermolecular energies. In all cases, the experimental crystal structures are close to the top of the lattice energy ranking. Phonon calculations on some of the lowest energy structures were also performed with DFTB3-D3 methods to calculate the vibrational component of the Helmholtz free energy, providing further insights into the solid-state behaviour of the target molecules. We conclude that DFTB3-D3 is a cost-effective method for optimising flexible molecules, bridging the gap between the approximate methods used in CSP searches for generating crystal structures and more accurate methods required in the final energy ranking.

A decomposition of Legendre polynomials into propagating angular waves is derived with the aid of an amplitude-phase method. This decomposition is compared with the ’Nussenzveig/Fuller’ so called near/far-side decomposition of Legendre polynomials. The latter decomposition requires the Legendre function of the second kind. This is not the case with the amplitude-phase decomposition. Both representations have the same asymptotic expressions for large values of \((l+1/2)\sin \theta \), where l and \(\theta \) are the polynomial degree and the angle respectively. Furthermore, both components of both representations satisfy the Legendre differential equation. However, we show the two representations are not identical.

Determining the range of conformations that a flexible pharmaceutical-like molecule could plausibly adopt in a crystal structure is a key to successful crystal structure prediction (CSP) studies. We aim to use conformational information from the crystal structures in the Cambridge Structural Database (CSD) to facilitate this task. The conformations produced by the CSD Conformer Generator are reduced in number by considering the underlying rotamer distributions, an analysis of changes in molecular shape, and a minimal number of molecular ab initio calculations. This method is tested for five pharmaceutical-like molecules where an extensive CSP study has already been performed. The CSD informatics-derived set of crystal structure searches generates almost all the low-energy crystal structures previously found, including all experimental structures. The workflow effectively combines information on individual torsion angles and then eliminates the combinations that are too high in energy to be found in the solid state, reducing the resources needed to cover the solid-state conformational space of a molecule. This provides insights into how the low-energy solid-state and isolated-molecule conformations are related to the properties of the individual flexible torsion angles.

We analyze the energy landscape of the 6 th crystal structure prediction blind test targets with various ab inito and semiempirical methodologies.A new benchmark set of 59 crystal structures (termed POLY59) for testing quantum chemical methods based on the blind test target crystals is presented.We focus on different means to include London dispersion interactions within the density functional theory (DFT) framework.We show the impact of pair-wise dispersion corrections like the semi-empirical D2 scheme, the Tkatchenko-Scheffler TS method, and the density dependent dispersion correction dDsC.Recent methodological progress includes higher order contributions in both the many-body and multipole sense.We use the D3 correction with Axilrod-Teller-Muto type three-body contribution, the many body dispersion MBD, and the nonlocal van der Waals density functional vdW-DF2.The density functionals with D3 and MBD correction provide an energy ranking of the blind test polymorphs in excellent agreement with the experimentally found structures.As computationally less demanding method, we test our recently presented minimal basis Hartree-Fock method (HF-3c) and a density functional tight binding Hamiltonian (DFTB).Considering the speed-up of three to four orders of magnitudes, the energy ranking provided by the low-cost methods is very reasonable.We compare the computed geometries with the corresponding X-ray data where TPSS-D3 performs best.The importance of zero-point vibrational energy and thermal effects on crystal densities is highlighted.

This paper describes a novel way to use the structural information contained in the Cambridge Structural Database (CSD) to drive geometry optimization of organic molecules. We describe how CSD structural information is transformed into objective functions for gradient-based optimization to provide good quality geometries for a large variety of organic molecules. Performance is assessed by minimizing different sets of organic molecules reporting RMSD movements for bond lengths, valence angles, torsion angles, and heavy atom positions.

This analysis attempts to answer the question of whether similar molecules crystallize in a similar manner. An analysis of structures in the Cambridge Structural Database shows that the answer is yes - sometimes they do, particularly for single-component structures. However, one does need to define what we mean by similar in both cases. Building on this observation we then demonstrate how this correlation between shape similarity and packing similarity can be used to generate potential lattices for molecules with no known crystal structure. Simple intermolecular interaction potentials can be used to minimize these potential lattices. Finally we discuss the many limitations of this approach.