Autumn 2017 Meeting
UKQSAR will bring the second meeting to the Manchester Institute of Biotechnology (MIB) facilities near the city centre in Manchester. The one-day meeting on the 29th September brings together representatives from big pharma, biotech and the academic communities to discuss a range of aspects reflecting on conformational analysis and prediction as well as big data analysis.
Conformational information of small molecule ligands is of central importance for rational drug discovery. Historically, conformational analysis was largely explored in silico using a range of methods. These approaches are now enhanced by the creative use information from experimental sources, principally NMR and small molecule x-Ray. Moreover, experimental methods offer a rich source of data and information which can drive the optimisation, interpretation and development of predictive methods.
In addition, new approaches are emerging which explore fundamental process leading to conformational selection, which limit the diversity of conformational space explored by small molecules. These efforts are of vital importance as experimental methods of conformational analysis suggest that the sampling of conformational space is more limited than predictive methods suggest. Thus understanding the impact factors which drive conformational behaviour is of central importance to inform conformational design efforts in computational and medicinal chemistry aimed at purposeful manipulation of the torsional behaviour.
The combination of enhanced conformational analysis and predictive methods with detailed analysis of “big” datasets of efficacy and non-efficacy data may offer an opportunity to meet the substantial challenge the pharma industry faces, reduction of attrition whilst significantly increasing efficiency and effectiveness in drug discovery.
With this in mind the programme as for the UKQSAR meeting was arranged to address a number of the issues raised and stimulate discussion with a view to continue the development of creative approaches towards new experimental approaches and the translation of these data into new workflows and tools.
About C4X Discovey
C4X Discovery exploits cutting edge technologies to design and create best-in-class small-molecule candidates targeting a range of high value therapeutic areas.
They have a state-of-the-art suite of proprietary technologies across the drug discovery process. The company’s innovative DNA-based target identification platform (Taxonomy3®) utilises human genetic datasets to identify novel patient-specific targets leading to greater discovery productivity and increased probability of clinical success. This is complemented by C4XD’s novel drug design platform which comprises two innovative chemistry technologies, Conformetrix and Molplex, that combine 4D molecular shape analyses (based on experimental data) with best-in-class computational chemistry. This provides new and unprecedented insight into the behaviour of drug molecules, enabling the production of potent selective compounds faster and more cost effectively than the industry standard.
Construction of the Manchester Institute of Biotechnology began in 2003 and was completed in early 2006. The architecture of the MIB reflects the needs of interdisciplinary science, featuring open-plan, multifunctional laboratories and a wide range of high-tech facilities as well as generously proportioned meeting and atrium areas to promote interaction, spontaneous discussion and shared research. It provides state-of-the art research and support space for approximately 600 research staff and up to 75 research groups over five floors. The Manchester Institute of Biotechnology was made possible by funding from the Wellcome Trust and the Wolfson Foundation.
Joint MGMS/UKQSAR Meeting, Spring 2018, Cardiff
Steve Maginn, Chemical Computing Group Inc
The 2018 Spring meeting of the UK QSAR group is being organised jointly with the Molecular Graphics and Modelling Society (MGMS) and will take place on April 11th and 12th at the University of Cardiff. The two day format follows that of the successful joint meeting held in 2016 with the PhysChem Forum.
Sessions are planned on “SAR Exploration and Prediction”, “Structure-Activity Relationships in the Solid State”, “Machine Learning and Other New Modelling Techniques” and “Water’s Role in Defining and Disrupting SAR”. Confirmed speakers so far include;
Ulf Norinder (Swetox, Sodertalje, Sweden)
Nora Aptula (Unilever, UK)
Bob Clark (Simulations Plus, USA)
Chris de Graaf (VU Amsterdam, Netherlands)
Sebastien Guesné (LHASA, UK)
Andreas Bender (University of Cambridge, UK)
Jonathan Essex (University of Southampton, UK)
Robert Docherty (Pfizer, UK)
Pedro Ballester (Centre de Recherches en Cancérologie de Marseille, France)
Ben Tehan (Heptares Therapeutics)
Graeme Day (University of Southampton, UK)
Alex Gaunt (Microsoft Research, Cambridge)
Registration will be free of charge, although the optional conference dinner, planned for the evening of April 11th, will be charged for. There will also be a commercial exhibition associated with the event, and a poster session. Potential exhibitors are invited to contact Steve Maginn at firstname.lastname@example.org . Registration and poster submission will open later this year.
We’re looking forward to having a wonderful meeting in Cardiff! Rydym yn edrych ymlaen at eich cyfarfod yng Caerdydd!
Talk Titles, Abstracts & Suggested Pre-Reading List for the Autumn 2017 Meeting
Title: Dissecting Interactions in Solution
Abstract: The study of non-covalent interactions is often complicated by their fundamentally weak nature and the role of solvent effects. This talk will illustrate how synthetic molecular torsion balances-- and supramolecular complexes can be used to quantify difficult-to-study interactions including van der Waals dispersion forces, aromatic stacking, hydrogen-bonding cooperativity, solvent,- and solvophobic effects. I will show how the information gleaned from such studies can be used to develop our theoretical understanding of molecular recognition phenomenon.
Jason Cole (CCDC)
Title: Analyzing molecular conformations using the Cambridge Structural Database
Abstract: The Cambridge Structural Database (CSD) contains over 900,000 crystallographically determined organic and metallo-organic small molecule structures. This vast amount of structural data provides a basis to analyze new structures, both experimental and theoretical, and to generate realistic structures for compounds of interest in a wide variety of fields including drug discovery and materials design. The Mogul application uses data from the CSD to derive frequency distributions for bond lengths, bond angles, and torsion angles and then compares a query molecule’s geometry against these distributions to flag cases of unusual geometry. Ring geometries are also considered by identifying similar rings to those in the query molecule in the CSD and comparing the torsion angles of the bonds in these rings to the query. As an extension to analyzing existing structures, these same knowledge-based distributions can be used to drive a method to generate a reasonable conformation of a new molecule. In this presentation, the effect of using this data in conformer generation and structural validation will be explored. Some recent extensions to Mogul that facilitate better treatment of rings will be discussed along with possible challenges that an end user faces in using solid-state derived distributions.
Martin Packer (AstraZeneca UK)
Title: Conformational analysis in molecular design – combining experimental and theoretical approaches for maximal impact
Abstract: Ligand conformational analysis can make a decisive contribution in molecular design. Using conformational analysis, we can develop hypotheses to enhance binding affinity or explain adverse structure-activity relationships (SAR). Experimental insights from NMR and X-ray provide us with the foundation on which we build, but design of novel compounds must rely on computational models. I will explore the extent to which experiment and theory concur in conformational analysis, across a range of ligand types and design hypotheses. The extent to which SAR meets design expectations is often controlled by fine details which computational models fail to predict or even explain. Experimental analysis helps to magnify the “small-print” of conformational analysis, revealing conformational preferences at the level of individual bond and atom types. Computational models provide a much broader brush for analysis and consequently are much less diagnostic of the fine detail of SAR. I will look at how best to account for weaknesses inherent in molecular models, so that we can aspire to maximise impact in design through a balanced combination of detailed “small print” from experiment with the “broad brush” pictures provided by modelling.
Matthew Habgood (Evotec UK)
Title: Comparison of methods for in silico generation of solution-phase conformational ensembles
Abstract: Knowledge of a ligand’s solution-phase conformational ensemble allows an additional level of rational design in pharmaceutical development. A recently developed technique for extracting conformations directly from NMR data now allows for a direct comparison between computational conformer generation and the real ensembles.
This talk will discuss the quality of ensemble reproduction achieved by three widely used conformer generation tools (MOE’s Low Mode MD, Xedex, and the CCDC’s conformer generator) over a test set of druglike molecules. Based on recent results in generation of extended versus folded conformers, explicit and implicit solvent molecular dynamics in Amber are also tested for comparison. All experimental ensembles are directly derived from NMR data.
It is concluded that explicit solvent molecular dynamics offers the best reproduction of NMR-derived solution phase conformational ensembles, although at vastly increased computational cost relative to dedicated conformer generation codes.
 C.D. Blundell et al, Bioorg. Med. Chem, vol. 21 (2013), p4976
 N. Foloppe et al, Bioorg. Med. Chem, vol. 24 (2016), p2159
Title: Learning Medicinal Chemistry ADMET rules from Cross-company Matched Molecular Pairs Analysis
Abstract: The first large scale analysis of in vitro ADMET data shared across multiple major Pharma has been performed. Using advanced matched molecular pair analysis (MMPA), we combined data from three pharmaceutical companies and generated ADMET rules, avoiding the need to disclose the full chemical structures and enabling pre-competitive learning. For all companies involved, in addition to the very large exchange of knowledge, there is a synergistic gain of approximately a fifth more rules created from the shared transformations than from just summing up rules from the individual companies.
There is good quantitative agreement between the rules based on shared data compared to both individual companies’ rules and rules published in the literature. Known correlations between logD, solubility, in vitro clearance and plasma protein binding also hold in transformation space, but there are also interesting exceptions. Data pools as large as this one allow focusing on particular functional groups and characterizing their ADMET profile. Finally the role of a corpus of robustly tested medicinal chemistry knowledge in the training of medicinal chemistry is discussed.
- J. Griffen, Macclesfield, England, S. Montague, Macclesfield, England, A. G. Leach, Macclesfield, England, A. G. Dossetter, Macclesfield, England, C. Kramer, Basel, Switzerland, J. Hert, Basel, Switzerland, T. Schindler, Basel, Switzerland, M. Stahl, Basel, Switzerland, A. Ting, Cambridge, UK, G. Robb, Cambridge, UK, H. Zheng, San Francisco, USA, J. J. Crawford, San Francisco, USA, J. Blaney, San Francisco, USA.
Title: The First Proteochemometric Model for the Bromodomain Family: Insights into Selectivity
Abstract: Recent efforts are beginning to deliver selective probe molecules for bromodomain proteins to elucidate key roles in oncological, cardiovascular and immuno-inflammatory disorders at the individual protein level.  Despite this, the design of selective bromodomain inhibitors remains a challenge. We present here the first modelling of bromodomain bioactivity data using global proteochemometric (PCM) modelling by integrating data made available at AstraZeneca and public data. The dataset has bioactivity values for 29 bromodomain targets, including 7 out of the 8 bromodomain subfamilies. The study aims to understand the key features which contribute to selectivity across the protein family and to predict the activities and selectivity profiles of new small molecules.
Generation of small molecule descriptors and sequence alignment-based descriptors of the protein active site has afforded the training of a highly predictive random forest classification model when validated on an external test set (balanced accuracy = 0.85, ROC AUC = 0.96). Furthermore, conformal predictions were used to provide an applicability domain for the models which were also validated using leave-one-group-out analyses and benchmarked against QSAR methods. Feature importance is used to highlight key binding site residues and their properties presenting as selectivity “hotspots”, which have been validated with literature observations. The model is being experimentally validated.
 Ferri, E.; Petosa, C.; McKenna, C. E. Biochem. Pharmacol. 2016, 106, 1–18. & http://pubs.rsc.org/en/content/articlelanding/2015/md/c4md00216d#!divAbstract
Prof. Jonathan Essex (Uni Southampton)
Title: Ligand pre-organisation – how well can computer simulations reproduce experimental conformational ensembles?
Abstract: In this presentation we will report the simulation and analysis of the solution-phase conformational ensembles of four inhibitors of the human fatty acid synthase. The ensembles are generated using replica exchange enhanced sampling molecular dynamics approaches for two force fields, and analysed using a combination of dihedral and Cartesian space clustering, and principal components analysis. These ensembles are compared to experimental data derived using NMR by C4X. We find that while the simulations are able to identify all the conformations found by NMR, their relative populations are in less satisfactory agreement. The use of these ensembles in terms of optimising ligand binding will be discussed
Title: The importance of NMR driven experimental conformational analysis in Drug Design