#EUGM10
Abstracts
Plenary Session
Frank Brown
Accelrys
Unlocking the Science – New Product Innovations from Accelrys
In this session, Accelrys executives will share the company vision and integrated roadmap and discuss how our merged company can help meet the challenges facing today’s scientific enterprise. Customers will learn about the plans for existing products and how our portfolio integration plans creates leverage across existing investments Accelrys and Symyx products and supports the evolving needs of today’s electronic laboratory environment.
Acclerys Technology Vision
A customer favorite at every User Group Meeting, this highly interactive session led by Matt Hahn,
CTO, will provide atendees with a preview of upcoming technologies currently planned or in development.
Experiences in Collaborative Development and Deployment of a Registration System for Biologics
The systematic determination of structural uniqueness and assignment of corporate identifiers to small molecules has been standard practice in the pharmaceutical industry for decades. This process is a requisite step for tracking inventory and serves to protect intellectual property, enhance communication and prevent duplication of effort. With therapeutically active proteins and other biologics becoming increasingly important to the industry’s bottom line, so too has the requirement to register these entities. The additional complexities involved with biologics, however, preclude their accurate and complete representation in typical small molecule registration systems.
As part of a Special Interest Group (SIG) consisting of Accelrys and a small group of pharmaceutical partners, Abbott assisted in the design and development of a system that properly registers biological entities, utilizing entity-appropriate business rules to define uniqueness. The Accelrys Biological Registration system is a Java and Oracle application that runs within Pipeline Pilot; it was designed to be highly configurable and extensible to meet unique customer requirements. Abbott was the first partner to deploy the system into production, in September 2009.
This presentation will provide a high-level overview of the system’s capabilities with an emphasis on how the system has been implemented at Abbott; specific topics covered will include uniqueness rules, tracking entity lineage and relatedness to other entities, the migration of legacy data and the integration of this system with other R&D applications. Abbott’s perspective on the SIG model that was used for the collaborative design and development of the system will also be presented.
René de Ruiter - Shell, Netherlands
Julian Willmott - Accelrys
The Challenge of Laboratory Informatics
This presentation provides a new approach to addressing some common challenges presented when implementing laboratory informatics solutions. The challenges discussed include having a broad system scope (spanning multiple departments and activities), the requirement to represent flexible laboratory workflows, the need to track relationships between items flowing through laboratories and the reality that instruments spew data in many different formats and locations. These are illustrated by a recent project between Shell and Accelrys where Pipeline Pilot was used to build a knowledge management system for their catalyst research and development business.
Accelrys Product Review Session
A customer favorite at every user meeting, this session continues the morning roadmap session with a discussion of the architectural and technology vision. We will preview some of the early integration work between several products in the portfolio and our product managers will join John to demonstrate some key upcoming releases. This session illustrates the integrated portfolio in action and highlights the many R&D organizations can leverage these solutions to streamline costs and increase their innovation productivity.
Pipeline Pilot Track
Visualising Multi-Dimensional Data Using Interactive Heatmaps To Identify New Drug Repositioning Opportunities
Using information analytics and Pipeline Pilot, we have built a robust platform for Drug Repositioning within AstraZeneca. Here we also detail a new interactive Heatmap component to enable end users to view multiple planes of data at the same time. This has enabled us to overlay information about our own compounds with information from scientific documents, competitive intelligence and manual curation. Not only does this tool offer us the ability to capture our knowledge about a potential new mechanism for New Opportunities to consider, but it acts as a single repository for the wider AstraZeneca scientific community to suggest new ideas and in the longer term offers exciting opportunities to consider ‘Open Innovation’ and provide a mechanism for interacting with scientists from across the world.
Pipeline Pilot implementation of an exhaustive hit expansion strategy combining bioisosteric transformations, similarity searches, and generalized substructure searches
Information about known active compounds and data generated in high throughput screening campaigns provide a valuable starting point for exploring and expanding the known active areas in chemical space. There is an abundance of chemoinformatics-based methods that allow the chemical space around active compounds to be explored, probing various similarity metrics, and determining molecular commonalities which govern activity.
Here we show that the combination of such approaches enables the detection of active compounds which could not be found using simple one-step methods such as fingerprint-based similarity methods. For example, bioisosteric transformations using bioisostere databases can be used for generating virtual query molecules which can then be subjected to similarity and generalized substructure searches in a second step. These algorithms, implemented in a Pipeline Pilot protocol, were applied to data sets compiled from the ChemBL database, demonstrating the suitability of such approaches, which can complement standard methods for more efficiently expanding the active chemical space and providing rapid access to SAR information.
Luca Sartori
IEO, European Institute of Oncology
Yet Another Registration System? Creating an Integrated System in a Changing World
Since middle of ’60s Compounds and Data Management Systems are some of the critical aspects of Chemical Informatics, especially in Pharmaceutical Industry.
Needs, constraints, problems and issues – on a relative limited scale– are similar when going from Academia and not-for-profit Research Institution to Biotech’s and large Pharmaceutical Companies. Hardware, software and manware complexities in small groups are the same as for big Pharma’s, and in some cases even worse. While recent trends in pharmaceutical industry point toward “In House developed” projects (examples are J&J and GSK), this is not possible for small groups, where “Manpower Critical Mass” has always been one of the limiting parameters. Therefore, in these last cases, the need for flexible, almost “off the shelf” but easily customizable and cost-effective applications is still acute.
At the same time the quest for a widely acceptable Chemoinformatic data model (and its corresponding User Interface) is still open. For example, the MDL (then Symyx) Default Registration Data Model was well suited to handle all the situations arising from initial steps of Drug Discovery, up to Development formulations and mixture handling. Nevertheless, some issues de-prioritized its further development. We’ll present the alternative developed to build up a Massive Registration System based on both Pipeline Pilot and Oracle PL/SQL technology coupled with the above mentioned Default Registration Data Model, to handle up to hundreds of thousands SDFile structure registrations.
The presentation will also briefly cover usage of Pipeline Pilot and Oracle PL/SQL as “glue applications” development tool, to provide different access to various databases and tools including DS Chemistry as well as examples of integration required in the field of Laboratory Automation (Electronic Lab Notebook, Assay Explorer & Corporate Database) and with the Compound Collection Sample Handling.
Obviously the merger of Companies creates big opportunities but, at the same time, a threat on existing applications that may disappear, thus severely impacting the whole Chemoinformatic workflow.
Frank Oellien
Intervet
Intervet Chemicals Directory (ICD) - A Framework Combining Accelrys Pipeline Pilot & Symyx Isentris
We present Intervet’s Chemicals Directory (ICD) a structural database that has become a key technology in the Drug Discovery process of Intervet Schering-Plough Animal Health. Formerly designed as source for all in silico approaches of the BioChemInformatics group, the ICD has become a central data source for many other groups like medicinal chemists and Compound Logistics today. In addition, contents like physico-chemical properties provided by the ICD play a critical role in decision support of many projects.
The ICD contains almost 11 million unique structures merging all compounds of Intervet Schering-Plough Animal Health in house database and more than 250 screening catalogues of 55 compound suppliers. The most important in silico properties like clogP, clogD, number of rotatable bonds and additional customized properties are available for all compounds and are used by many other front-end applications like our Decision Support Framework.
As early adopters of MDLs Isentris technology (now Symyx), the ICD was one of the first implementations using Isentris technology in the industrial pharma field. Structures are stored in the Symyx Direct cartridge. Structure registration has been performed in the past by self-developed Java applications based on Symyx Isentris Client for Java including chemical normalization by Cheshire.
Unfortunately, the Isentris-based set-up has several limitations regarding automation and workflow management. To overcome these limitations we have used Pipeline Pilot as a complementary technology and in some scenarios as a full replacement for Isentris: Calculation and storage of all physico-chemical properties is done automatically by PP. Furthermore, PP workflows for the maintenance of the whole data pool have been developed leading to a massive decrease of time rescores. Finally, the Isentris-based synchronization of the in-house database and the ICD could be completely replaced by an automated PP workflow and a Java-based PP Cheshire component.
Implementation of Symyx ELN at Alere San Diego
One of the fastest growing markets in scientific informatics is that of Electronic Laboratory Notebooks (ELNs), which have progressed from the periphery of pharmaceutical research to playing a central role in research and development not only for the Life Sciences, but also the Chemical & Petrochemical and Biotech Industries over the past 4 years. Not only have they have made information accessible enterprise-wide but also provide primary evidence for the ownership of intellectual property. Where originally ELNs were implemented in Medicinal and Process Chemistry they are now moving into the biological sector, device and diagnostic companies, where efficiency gains of 15–25%— about four times that of medicinal chemistry is possible [1]. This presentation will describe our wonderful journey towards Symyx ELN at Biosite, Inverness Medical Innovations Inc., and Alere San Diego, a diagnostic companies, and beyond.
[1] Keith T Taylor, “The status of electronic notebooks for chemistry and biology” Current Opinion in Drug Discovery & Development 2006, Vol 9 No 3, p348-353.
Richard Carter
Oxford Nanopore Technologies
Novel Applications for Next Generation Sequencing made possible with Pipeline Pilot
High throughput next-generation sequencing (NGS) produces large amounts of data in short periods of time, allowing users to study genomics on an unprecedented scale. Data acquisition rates are growing enormously and the genomics community is also anticipating new technology generations. There is a growing need for accessible and effective data handling tools for both large and small users. Oxford Nanopore Technologies, developers of a new technology platform for direct, electronic detection and analysis of single molecules, has assembled compelling new functionality with Pipeline Pilot’s new NGS collection. Using years of data analysis experience at Solexa, Illumina and Oxford Nanopore, typical workflows have been designed that will help users produce biologically relevant information and integrate with other components and collections in Pipeline Pilot.
Ed Ainscow (AstraZeneca)
A Flexible Platform for Processing High-Content Screening Data
High-content assays produce large volumes of multi-parametric data which present a challenge to any data analysis tools. This talk presents a case study of the use of Pipeline Pilot to address the data handling issues of a High-Content assay development group. This group not only faces the normal challenges of dealing with High-Content data, but also the challenges of assay development where new instruments and new assays are frequently introduced, and where the diversity of data formats represents an ever-changing challenge.
This talk will focus on how Pipeline Pilot has been used to implement a highly flexible platform for processing high-content data which provides a common, consistent framework, whilst easily allowing new types of data and new workflows to be seamlessly integrated. Using this platform, scientists can work with data from different instruments, imaging modalities and dimensionalities, and are empowered to develop not only the assay, but the data processing approach to accompany it.
Project Dashboard Integration with Pipeline Pilot
In past User Group Meetings there have been talks describing how Pipeline Pilot can be used to track the progress of the discovery pipeline in the Pharmaceutical industry. Unlike the Pharmaceutical industry, Unilever operates in areas as diverse as Ice-creams, shampoos and floor cleaners. In this case, tracking a project, or several projects cannot be focused on compounds and their progress through the pipeline but is based upon more generic attributes like safety and efficacy.
This presentation describes a Pipeline Pilot driven application that uses a combination of traditional web application, email, web forms and a database to obtain, structure and store monthly project status information across the full range of Unilever projects in a single consistent form. Once the data has been collected the summaries and differences are reported to Unilever senior management in a range of formats.
Andrea Beccari
Dompé
Multi-dimensional Hierarchical Scaffold Analysis
The analysis of chemical diversity of huge collections of commercially available, synthetically accessible, or in silico designed, compounds have become a topic of considerable interest in the drug discovery process to guide the exploration of chemistry space [1,2]. Also the problem of recognition and classification of meaningful relationships between shared chemical features and target biological activity, involves the analysis of large sets of diverse compounds, generally produced during high-throughput screening (HTS) campaigns.
To address these topics many different solutions have been developed in the last twenty years. The most relevant strategies can be divided in different groups: property based, structure based, or pharmacophoric based descriptors. Recently the scaffold based approaches have become widely implemented due to the efficiency in the clustering analysis and in the database indexing [3,4,5]. They are also generally more interpretable from a medicinal chemist prospective.
In this context we developed a new flexible algorithm that is able to generate different levels of scaffold’s abstractions (scaffolds, frameworks, and wireframes) and deconstructions. This procedure generates unbiased multi-dimensional hierarchies, which are effective tools in the HTS data analysis where, due to the diversity driven composition of the screening libraries, conventional scaffold representations often are not the most suitable metrics for compound clustering.
We present a complete and easy-to-use Pipeline Pilot’s collections of components that allow generate, manipulate, and view these scaffold’s abstractions and deconstructions.
Pharmacophore based virtual screening of non-enumerated combinatorial compound libraries.
Pharmacophore based virtual screening of compound libraries usually requires full compound structures and are not suitable to fragment screening. Hence, screening a fragment library requires a combinatorial enumeration of the fragments according to validated chemical reactions. This process may generate a huge number of compound structures and is computationally intensive. Furthermore, once enumerated, structures have to be compared to the pharmacophore and scored, yet another computationally intensive process.
We present a method to screen non-enumerated compound libraries using Pharmacophore Graph1. Fragments are compared to pharmacophore queries and scored and only those partially mapping the pharmacophore query become eligible to combinatorial enumeration. Fragment combination rejects fragment pairs overlapping onto the pharmacophore query. This method considerably reduces the number of enumerated structures and does not require further pharmacophore comparison of the enumerated structures.
1. Pharmacophore Graph is a 2D-pharmacophore method available in Pipeline Pilot as part of Discngine Chemistry Component & Protocol Collection – available from www.discngine.net
Brad Sherborne
Schering-Plough
Exploring Data without Forms - Breaking Moulds for SAR Delivery
Pipeline Pilot is an excellent platform for rapid implementation of novel (cheminformatics) algorithms. Following the web publication of an algorithm to identify molecular matched pairs (Hussain and Rea: JCIM 2010(50)339), a protocol to reproduce the method was implemented in a day.
Transforming this pilot and related protocols into robust and friendly applications for chemists to explore chemical structures and activity data will be illustrated along with ways to make exploring such landscapes visually and interactively engaging.
Will Pitt
UCB
Fun with Chemical Building Blocks: Building New Heteroaromatic Ring Systems with Pipeline Pilot
It is now well understood that the creation of a complete list of all possible drug-like molecules is not possible. However, if such molecules are thought of as being made up of discrete components, then a deconstructed chemical universe can be studied in its entirety. The creation of theoretically complete inventories of these chemical parts has advantages that go beyond basic stock taking. Lists of this sort, when linked to the chemical literature, can also be used to provide a source of inspiration for synthetic chemists. It is possible to combined known chemical fragments in new ways to produce completely novel chemical moieties. The focus of this talk will be the prediction of the structures of synthetically tractable, yet completely unknown heteroaromatic ring systems. All ring systems in the complete set (VEHICLe) are found to be made up of just 346 parts (ECFP_2 fingerprint bits). However, synthetic chemistry is a “long-tailed” science. Some chemical components are extremely common while many more are extremely rare. The implications of this fact for the prediction chemical tractability and for the size of the most probable chemical universe will be discussed. How Pipeline Pilot was used to construct VEHICLe and to predict synthetic tractability will be described. The handling of tautomeric equivalents will also be mentioned.
Stephen Pickett
GlaxoSmithKline
Development of an Automated QSAR Workbench with Pipeline Pilot
GSK has had a long term interest in the development of a robust QSAR platform to support both the specialists in the development of globally applicable models, in the area of ADMET for instance, and to provide validated approaches for computational chemists involved in programme support [1]. The Automated Modelling Environment [2] was developed in-house in response to this need. Changes in the organisational landscape meant that support for this system was too heavy a burden on our resources and hence we decided to explore the potential for Pipeline Pilot as an alternative solution. In this presentation we shall discuss how we took the AME concept and with appropriate Accelrys resource were able to move first to a prototype in the style of the MedChem Workbench and subsequently to a fully fledged QSARWB within less than 12 months. The QSARWB provides the functionality we require for building, validating, reporting and deploying models globally whilst giving us the ability to customise and utilise our own science through the use of the standard Pipeline Pilot protocols. Importantly, the support overheads are lower.
[1] Stephen Pickett, EuroQSAR 2006
[2] Darren Green, http://www.soci.org/News/Fine-Chemoinformatics-SAR
Predictive Models for Key Compound Quality Indicators That Stand the Test Time
- How can we ensure our computational models have the best accuracy and precision?
- If we can automate the building of these models - would we see a benefit over handcrafted models from a computational “artisan” ?
- Would medicinal chemists ever trust a system automatically building and maintaining 1000s of qsar models/week ?
- What would the potential impact of such a system be?
- How would this transform the role of computational chemists?
Christophe Buyck
Johnson & Johnson
The use of the Biologically Directed Diverse Library (BDDL) approach to select compound subsets out of vendor offerings - a comparison with a travel agency
Selecting a set of compounds out of an offering of a vendor can be achieved in many ways. Here we describe a selection procedure that tries to sample a diverse subset with a bias towards compounds that would have a higher propensity of being biologically active. The talk will cover both the procedure used to capture this biological relevant chemistry space as well as how to subsequently select the compounds out of this set.
Discovery Studio Track
Announcing the Release of Discovery Studio 3.0: an Improved Molecular Modeling Environment to Do More
In 2010 the pressure to do more with less continues to be the driving theme in industry, government and academia. Researchers are being asked to optimized their projects, lower the number of experiments and overall cut cost for research and development of new molecular entities. In order to assist with this challenging task, applications like Discovery Studio and Pipeline Pilot continue to evolve to help the researcher get their answers faster and easier. In this talk, we will be highlighting various improvements made to our flagship modeling and simulations environment, Discovery Studio, in the context of Performance, Usability and New Science. Though the main theme for Discovery Studio 3.0 release is in the area of Biological Design and Engineering, enhancements have been made across the program including Pharmacophore improvements, automated typing, protein stability prediction, antibody automation, improved integration to Pipeline Pilot and much more.
Structure of IL-15 in Complex with a Potent Human Neutralising Antibody Reveals Heavy Chain CDR3 alpha-helix as a Key Component of Paratope
IL-15 is an inflammatory cytokine that has been implicated in mediating inflammatory diseases.
A large panel of human neutralising anti-IL-15 antibodies have been isolated from the MedImmune scFv phage display library. Subsequent affinity maturation of a selected antibody by targeted mutagenesis of the heavy and light chain CDR3’s has generated a potent lead (DISCO 280) for the potential treatment of autoimmune diseases such as rheumatoid arthritis.
The co-crystal structure of DISCO 280 Fab with IL-15 has been solved, revealing the binding epitope of the antibody. It reveals structural insights relevant to antibody optimisation and engineering. The paratope is characterised by an alpha helix in heavy chain CDR3. Interestingly some the changes introduced during lead optimisation are not directly part of the antibody-antigen paratope but rather involved in heavy and light chain stabilisations of the antibody Fab fragment.
Our analyses suggest that even small changes in the antibody-antigen interface provide large improvements in the free energy of antibody-antigen binding. Furthermore, the analysis is consistent with the optimisation process introducing amino acid residues which lie outside of the direct antibody-antigen contacts but which contribute to the improved affinity of the antibody.
Designing Biologically Relevant Isosteres using FieldStere in Discovery Studio
We have developed a unique ligand based drug design methodology based on molecular interaction fields. Previously we have described the use of our Field technology in molecular alignment, virtual screening and pharmacophore perception. A recent extension to this technology facilitates the searching of fragment databases for biologically relevant replacements for parts of a molecule. In the FieldStere approach, a substructure of a known active is marked for replacement; a database of fragments is then searched for moieties that are isosteric and isoelectronic. Uniquely, replacements are scored in “product” space as final molecules allowing for electronic and steric influence between replacement fragment and retained sections of the potential inhibitor.
In this presentation we will describe the use of Discovery Studio and Pipeline Pilot to access the FieldStere technology. Using a live demonstration we will show how you can easily find and define FieldStere search queries and then use these to generate biologically relevant isosteres. Using the flexibility of Pipeline Pilot we will show how the bioisosteres can be further evaluated and refined, resulting in a short, high quality hit list for synthetic follow-up.
Computational Biology Applied to Early Development of Antibody Therapeutics: Key Aspects of Advancing Antibody Fragments from Research to Clinic
Single chain Fv antibody fragments (scFvs) were developed more than 30 years ago. Despite their favourable pharmacokinetic properties, scFvs are not commonly used as therapeutics mainly due to low stability and tendency to aggregate. ESBATech’s antibody platform very rapidly delivers highly-stable scFvs to be developed as therapeutics for a wide range of targets.
Our technology is based on generation of high-affinity antibodies by immunization. Since lengthy and unpredictable in-vitro affinity maturation is not required, this platform generates panels of high-affinity antibodies in only 8 weeks. Here we will discuss the importance of computational biology with respect to the antibody lead selection process. In addition, we will address the importance of using in-silico models in the design and optimization of highly-stable scFv molecules and how these insights facilitate the preclinical development of antibody drugs.
Annalisa Nuccitelli
Novartis Vaccines
Discovery Studio: A Powerful Tool for Vaccine Design
Despite evident progresses in generating efficacious treatments against many infectious diseases, bacterial pathogens remain the most important threat to health worldwide. For this reason, new vaccines and methods to improve their development are of major interest.
In this context, Discovery Studio resulted to be a very powerful tool for structural characterization of potential vaccine candidates allowing us to optimize stability, homogeneity and easiness of production.
The Homology Modeling suite available in Discovery Studio helped us to rapidly generate models of the vaccine candidates which helped us to identify of the immunogenic domain of each vaccine candidate. Moreover, the sequence analysis tools (Phylogenetic & Evolutionary Trace Analysis) allowed us to identify and visualize onto the 3D structure crucial residues for each protein family leading to the rationalization of functional mutagenesis.
Didier Rognan
Université Louis Pasteur
Massive Target Profiling by Parallel Protein-ligand Based Pharmacophore Searches
Target profiling is an efficient way to control the selectivity of bioactive compounds, prioritize hit-to-lead development or lead optimization, predict off-targets and therefore unwanted side-effects. By opposition to major pharmaceutical companies, most academic research centers lack the necessary manpower, infrastructure and economic resources to address this issue at the experimental level. Developing computational methods to predict the binding of a ligand to a wide panel of targets [1] is therefore an intense area of research. Several approaches are possible ranging from pure ligand-based similarity searches [2], pharmacophore mappings [3], protein-ligand docking [4] to protein-ligand active site comparisons [5]. We herewith present an attempt to mine the protein-ligand chemogenomic space covered by the sc-PDB dataset of 7,078 druggable protein-ligand complexes [6] by i) automatically generating pharmacophore queries from high-resolution protein-ligand X-ray structures, ii) systematically mapping a set of 3,687 PDB ligands to > 50,000 pharmacophore models. Different ways of mining this matrix of ca. 200 billion data points will be presented and strengths/weaknesses of the method presented.
1. Rognan, D. (2010) Structure-based approaches to target fishing and ligand profiling. Mol. Inf., 29, 176-187.
2. Keiser, M.J. et al. (2009) Predicting new molecular targets for known drugs. Nature, 462, 175-181. 3. Steindl; T.M. et al. (2006) Parallel screening: a novel concept in pharmacophore modeling and virtual screening.J. Chem. Inf. Model. ,46, 2146-2157.
4.
Müller, P. et al. (2006) (2006) In silico-guided target identification of a scaffold-focused library: 1,3,5-Triazepane-2,6-diones as novel phospholipase A2 inhibitors. J. Med. Chem., 49, 6768-6778.
5.
Kinnings, S.L. (2009) Drug discovery using chemical systems biology: repositioning the safe medicine Comtan to treat multi-drug and extensively drug resistant tuberculosis. PLoS Comput. Biol., 5, e1000423.
6.
Kellenberger, E., Müller, P., Schalon, C., Bret, G. and Rognan, D. (2006) sc-PDB: An annotated database of druggable binding sites from the Protein Data Bank. J. Chem. Inf. Model., 46, 717-727.
Rod Hubbard
University of York – YSBL – Vernalis
Fragmentology
The past ten years has seen tremendous developments in the experimental methods of “Fragment-Based Lead Discovery”, FBLD. The central feature is that the drug discovery process begins with identification of small (<250 MW), weakly binding (affinity of 100s of µM) compounds which are then optimised to drug candidates by structure-guided design. The advantages are that a small library can sample a potentially large chemical diversity to generate novel lead compounds and that hits can be identified for new classes of target for which existing compound collections cannot provide a hit. The main challenges are design of the library, robust identification of which fragments bind and the need for structural information to decide how and which fragments to progress.
The presentation will begin with a brief summary of the key features of current practise in FBLD, illustrated with some projects at Vernalis that have successfully delivered clinical candidates.
The core of the presentation will be on how computational and cheminformatics methods can and should be supporting this new era of fragment-based discovery. Ironically, it was the computational chemists and modellers who made the first forays into both structure and fragment-based discovery in the early 1990s. Perhaps that early work was premature (i.e. it didnt work), but there is now the experimental data and imperative to revisit, refine and validate the methods.
I will present recent work from York in two areas - fragment library design (using Pipeline Pilot protocols) and prediction of fragment binding poses (MCSS within Discovery Studio).
I will conclude with some thoughts on the challenges and opportunities provided by the growing confidence and investment in structure and fragment-based discovery. There is a real need (and experimental data against which to develop the methods) in areas such as de novo design (linking, merging, growing fragments), synthetic tractability (to assess such design but also to assist in choosing which fragments to progress) and the growing database of structural, thermodynamic and kinetic data being generated for protein-ligand complexes.
The work on MCSS is a collaboration between a York student, Kamran Haider and Hugues-Olivier Bertrand of Accelrys. The fragment library design is by York students Michele Schulz with some assistance from Isaac Bruce.
Recent advances in GOLD and its implementation within modelling workflows via Discovery Studio and Pipeline Pilot
GOLD remains one of the most powerful and widely used protein/ligand docking packages worldwide and there is a strong commitment within CCDC to continually improve its functionality and practicality of use. Recently we have added Ensemble docking, which allows the user to efficiently dock against multiple models of the target protein, thus taking account of protein flexibility. We have also developed new scoring functions which, either on their own or in combination with existing scoring functions, can significantly improve pose prediction and virtual screening success.
These recent advances will be described. We will also describe how an active collaboration between Accelrys and CCDC is making it much easier to incorporate GOLD docking into modelling workflows involving Discovery Studio or Pipeline Pilot.
Anna Tempczyk-Russell
Accelrys
Ligand based modeling studies – Novel and Diverse Ceramide Glucosyltransferase Inhibitors Identified from the ACD Search
Ceramide Glucosyltransferase enzyme catalyzes the first glycosylation step in glycosphingolipid biosynthesis. Transfer of glucose to ceramide is an important step in a variety of cellular processes such as cell recognition, growth, development and cell differentiation. Inhibition of this enzyme can have therapeutic relevance for multiple disorders. We have analyzed experimentally assayed compounds to calculate their chemical and structural properties to correlate them with measured activities. In order to understand the mechanism of inhibition for each compound we measured its similarity to the substrates, products or the intermediates of the catalyzed reaction. Comprehensive analysis of all key contributors to the mechanism and potency of the inhibition lead to the definition of 3D pharmacophores. These pharmacophores were used to search the ACD database. This search resulted in the identification of novel, chemically diverse potential inhibitors outside of the patent space.
Rachel Clark,
Strathclyde University
The Drug Discovery Portal’s Approach to Exploiting the Biological Activity of Natural Products
Natural products continue to offer benefits to drug discovery programmes despite them falling out of favour with industry. The chemical novelty of natural products is higher than that of any other source and their rigidity, chiral complexity and evolutionary development of intermolecular hydrogen bonds make them interesting leads for ligand based design.
The Drug Discovery Portal (DDP) at the University of Strathclyde is a collaborative tool to enhance drug discovery collaborations between chemists and biologists in academia. Biologists who are looking for compounds with specific biological activity can engage with our computational scientists to design a virtual screening assay to identify compounds from our proprietary chemistry collection or from commercial sources. Often there is insufficient structural data to initiate a structure based virtual screen, and instead we work with a ligand or series of ligands to build pharmacophore queries. In our experience, ligand based design using a complex natural product as a starting point yields hits of less complexity with a similar activity profileItai Bloch
Dynamix Pharmaceuticals
Efficient In-Silico Screening through Integration of Discovery Studio and Pipeline pilot with Database and a High Computing System
The goal of Virtual Screening (VS) is to enable rapid evaluation and filtering down of large numbers of chemical structures within a short period of time. The challenge is combining multiple computational tools to filter down extremely large compound collections into a small subset of selected compounds that can then be tested for biological activity, and doing so in a time- and resource-efficient way.
We have developed a general workflow to address this challenge and efficiently manage the vast amount of data generated during the VS process. Our workflow integrates different computational approaches, tools and systems into a seamless platform technology. This platform combines Discovery Studio protocols and home-grown algorithms together within the Pipeline Pilot framework, communicates with our underlying compound database and manages the traffic to and from the high performance parallel computing servers. As a result this platform enables rapid and efficient Virtual Screening. This method has proven powerful in rapidly advancing our discovery programs from idea to actual drug candidates, with validated activity in biological assays.
Specifically, among the components of this system we have modified the way compounds, and compound collections, are managed and tracked during the VS process. Following each analysis or filtering step, the selected compounds are tagged and iteratively uploaded into and retrieved from the underlying database. Direct and continuous communication with the database, rather than handling vast numbers of large individual files, saves both storage space and valuable time, resulting in a highly efficient virtual screening process and significantly reduces human errors. The efficient data management and recording system developed within this platform makes the data highly accessible for analysis and retrieval. Another key feature of this workflow is the integration between ligand-based and structure-based approaches, taking advantage of essential Structure Activity Relations data and 3D Structural insights.
Materials Studio Track
Richard Catlow
FRS, University College London
Advances in Computational Studies of Energy Materials
This lecture will review recent developments and applications of computational techniques in the field of materials for energy technologies, including hydrogen production and storage, energy storage and conversion and light absorption and emission (1). We will highlight recent work on tin titanate photocatalysts, mixed metal sulphide solar cell absorbers and the mechanisms of incorporation of hydrogen into silver and copper oxides. Special attention will be paid to the modelling of nano-structured systems, including ceria and group
13 sesquioxides. We consider applications based on both interatomic potential and electronic structure methodologies; and we illustrate the increasingly quantitative and predictive nature of modelling in this field.
(1) CRA Catlow et al., Phil Trans R Soc 368, 3379, (2010)
Prediction of Catalyst Properties using QSAR
In the field of homogeneous catalysis a lot of research is devoted to getting a better understanding of the relationship between activity/selectivity of catalysts and structural features of the involved complexes. In the cases of transition metal complexes bearing triarylphosphines this can be achieved by looking at linear relationships between activity/selectivity data and a known parameter. Examples include Hammett plots and plots of NMR coupling constants vs catalysis data. A major drawback is that these relationships are usually obtained afterwards, when the catalytic results are known.
We are interested in finding new methods to predict properties of new homogeneous catalysts by using QSAR techniques. Although QSAR is becoming more and more a tool to make large catalyst libraries in silico and pick the best candidates from those, it usually requires the input of your specific data set to generate QSAR models that can be applied only to your specific reaction. Using data from both literature and our own work we have been looking into generating QSAR models that do not predict new catalysts based on activity data but on more general properties like those stated above. In the presentation we present our current results obtained.
Catalyst Design & Development
Irrespective of whether computational or laboratory based experiments are performed the infrastructure required to allow the development of valuable knowledge bases lags behind the cutting edge tools used to generate the data. With the technology advances in high throughput experimental rigs and the increases in computer performance the data streams can be large in volume and density. Therefore the bottleneck has shifted to the interpretation of data and anything that can be done to aid researchers will increase productivity and allow more sustainable growth. Additionally, for global companies and collaborative projects there is a need to cater for multiple sites potentially on different continents. If data can be made easily accessible new possibilities for mining and interpretation become feasible.
Examples will be given highlighting how it is possible to improve on current standard practise and make more of the information we have.
Materials Studio 5.5: New Developments for Catalysis, Energy, and Functional Materials Applications
The latest version of Materials Studio contains many new enhancements. Whilst these are generally applicable to the Chemicals, Materials, and Pharmaceutical industries, they have been targeted at functional materials such as opto-electronic films and coatings, battery, and hydrogen storage materials. This presentation will highlight some of the new features and functionality that will be available in Materials Studio 5.5 including CASTEP, DMol3, and Materials Visualizer.
Simulation of Laser Damage & Transient Absorption in Vitreous Silica
Femtosecond laser pulse induced structural changes in silica glass and their role in changing the refractive index of the glass have been investigated using ab initio molecular dynamics simulation. Femtosecond laser irradiation was simulated by raising the electron temperature to 25000 K and allowing the system (72 atom cell) to evolve freely for 300 fs. During the irradiation the average nearest-neighbor Si-O, Si-Si and O-O distances increase due to the weakening of bonds resulting from the thermalization of electrons. Diffusion of Si and O gives rise to a structure with 2- and 3-coordinated Si atoms in addition to non-bridging oxygens. These structural changes are almost completely recovered during post-irradiation evolution of the glass structure. However, there are persistent changes that involve the formation of three-coordinated Si atoms and non-bridging oxygens that correspond to the paramagnetic defect species of Si E centers and non-bridging oxygen hole centers, respectively. These defects introduce energy levels within the band gap of silica glass giving rise to optical absorptions that increase the refractive index through a Kramers-Kronig mechanism.
Transient absorption in silica is thought to be due to the formation of a new SiH species (SiH*, Smith et al., Appl. Optics, v. 39, 5778). A model to explain the results involves the reaction of H2 to produce a three coordinated oxygen (SiOHSi) and a 5-coordinated Si, with four Si-O bonds and one Si-H bond. This model predicts activation energies of H2 reaction similar to experimental values and also predicts the correct trends in the changes of absorption and Raman spectra upon irradiation.
Computer-Aided Materials Design in Adhesive Technologies
Computer-based simulations provide a valuable means for the prediction of chemical processes and improvement of material properties. Simulations as virtual experiments for the development of novel adhesives help to focus experimental effort. Thus, the overall product development cycle for new materials and active ingredients can be reduced significantly.
The method portfolio ranges from quantum chemical, molecular mechanics and mesoscopic simulations to QSAR approaches. Practical examples will be given how scientific computing methods contribute to innovative adhesives technologies both, for consumer goods and industrial applications.
From Lattice Vibrations to Thermodynamics
Neutron scattering and lattice dynamics have a long history, in particular with experimental data being used to refine inter-atomic force constants. Nowadays there is still a strong synergy between experiment and simulation, with the difference that ab initio computational methods are often now used to prepare experiments and analyse inelastic scattering data from complex systems. In addition, ab initio methods give access to electronic and magnetic structure, the latter being directly probed by neutron scattering techniques.
I will therefore show a range of examples of how neutrons and numerical methods are giving clear insights into the role of structural excitations selected from DNA [1], short hydrogen bonds [2], solid oxide fuel cells [3], thermoelectric materials [4], magnetic materials, including iron-pnictide superconductors [5], phase change memory materials [6].
[1] Merzel et al., Phys. Rev. E (2007) 76 31917
[2] Fontaine-Vive et al., J. Chem. Phys. (2006) 124 234503
[3] Koza et al., Nature Mater. (2008) 7 805
[4] Paulus et al., J. Am. Chem. Soc. (2008) 130 16080
[5] Zbiri et al., Phys. Rev. B (2009) 79 64511
[6] Otjacques et al., Phys. Rev. Lett. (2009) 103 245901
Evelyn Moreno
University of Barcelona
Aliphatic Materials: Crystal Structure Resolution, Polymorphism and Energy Storage Applications
Abstract to be provided
Simulating the PVT Behaviour of Polyamides: Polyamide 6 as a Test Case
PVT data are of interest to DSM because they are required for proper prediction of product properties and shape via simulation of injection moulding. For a number of (commercially) interesting engineering plastics reliable experimental data is lacking since samples have to be exposed to high temperature for a long time leading to serious thermal degradation. A typical example for which this is the case is DSM’s Stanyl, i.e. polyamide 46. Using polyamide 6 as a test case and applying a similar strategy as Rigby & Eichinger (Pol. Int. 44 (1997) 311-330) we recently investigated whether atomistic simulations could be a viable alternative for predicting the PVT behaviour of polyamides. Our strategy and most recent results will be presented.
Philippe Lienard
Sanofi-Aventis
Predictive API Chemical Weakness Assessment using in silico Approach Combined with Chemical Degradation Based Knowledge
The benefits of evaluating, early in the development, the active pharmaceutical ingredient (API) candidates for their chemical stability properties are well recognized, as the pharmaceutical scientists need to design formulations that guaranty an acceptable and minimal chemical degradation of the drug product.
An in silico approach is combined with a chemical degradation knowledge expertise to help understanding and anticipating the stability of the APIs; it allows to advise the pharmacists for formulation design of the drug product to minimize chemical degradation. The used molecular modelling methods are based on the Fukui indices for anticipating nucleophilic/electrophilic degradation pathways like hydrolysis and electrophilic oxidation. In parallel, a hydrogen abstraction radical methodology is utilized to predict radical oxidation by oxygen of air, also called autoxidation.
Modeling of Crystal Growth Inhibitors
Abstract to be provided
Workflow Automation with the Materials Studio Collection
The Materials Studio Collection is a new software solution that allows access to Materials Studio's extensive materials modeling capabilities within Pipeline Pilot, Accelrys’ scientific informatics platform. The combination of Pipeline Pilot’s data management, reporting and analysis tools with Materials Studio offers a wealth of opportunities for scientists to rapidly develop and deploy new applications that automate the simulation of scientific or analytical processes and the prediction of properties , and to deploy these methods to a wider user base.
We will present a number of protocols and applications that demonstrate the power of this approach: from applications to automate a range of computational steps for large numbers of systems (high-throughput modelling) to the development of new scientific methods based on the existing toolbox of algorithms. We will conclude by presenting examples of current work under development to further enhance the Materials Studio Collection.
Gerhard Goldbeck-Wood
Accelrys
Bridging Length & Time from the Electronic to the Nanoscale: Results from the Accelrys Nanotech Consortium
Bridging length and time from the electronic to the nano-scale: results from the Accelrys Nanotech Consortium
The design of new materials and products at the nano-scale means that quantum and molecular simulations can provide key and direct insights. On the other hand, many phenomena still stretch time and length scales that are beyond the limits of many standard tools. Also, interfaces between different materials are often crucial, but are typically hard to model.
Since its beginning in 2004, the aim of the Nanotech Consortium has been to address some of these challenges, and create new opportunities for modelling to impact nanotechnology innovation. As the Consortium is drawing to a close after six years of collaboration with more than thirty leading organisations, the presentation will review how the challenges have been addressed with new methods now integrated into Materials Studio.
Thomas Keal
Science & Technology Facilities Council
Massively Parallel Materials Modelling with ChemShell
The ChemShell computational chemistry environment (available in Materials Studio as QMERA) offers a route for the efficient modelling of surface reactions by combining high-level quantum mechanical (QM) techniques for the reactive region with a cheaper molecular mechanical (MM) model of the solid environment.
There is increasing interest in carrying out very large scale calculations with ChemShell which require good parallel scaling performance. The current release of ChemShell can exploit the parallelism of the QM and MM energy evaluation programs but does not take advantage of potential parallelisation of higher level tasks (such as reaction path optimisation or finite-difference Hessian evaluation). We have developed a task-farming parallelisation framework for ChemShell which gives substantially improved performance on HPC systems for these tasks.
The new techniques have been used to study the heterolytic dissociation of hydrogen over a polar oxygen-terminated surface of aluminium-doped zinc oxide, using a 3200-atom QM/MM model. Forward and backward dissociation barriers are calculated using the nudged elastic band method. The results corroborate experimental evidence that the reaction is irreversible.
Modeling of the Fischer-Tropsch Process
Fischer-Tropsch is a process in which syngas (a mixture of CO and hydrogen gas) is converted into higher hydrocarbons. The method was developed during the 1920’s. The process has always received much attention when oil was expensive and in countries that could not get oil easily. The process is very complicated, as a very large number of hydrocarbons can be formed. Even after many decades of research some fundamental questions have still not received a definite answer. It is not known, for example, if the rate-determining step of dissociation of CO is direct or proceeds via CHO.
I will show how one can model such a complicated process using kinetic Monte Carlo. I will discuss various ways to model different sites in the unit cell, how to model steps, how to deal with lateral interactions, what to do when a fast equilibrium takes up all computer time, how to avoid jamming of the adlayer, and some ways to reduce the complexity of a reaction model.
A separate problem that I will discuss is how to obtain rate constants and lateral interactions that the kinetic Monte Carlo simulations need as input. I will briefly show how to get them from experiments, but I will mainly focus on the calculation of these properties using density-functional theory.
Workshops
IT Best Practices
The workshop will focus on best practices related to PP administration and development. We encourage the audience to discuss best practices in their organization.
Topics to include: Component Design, Protocol Design, Application Packaging, Production Deployment, Server Upgrades, Migration and Disaster Recovery, Protocol Regression System, Server Sizing and OS Selection, Server Configuration, Development, Integration, SharePoint Integration, Deploying Web Port Beyond the Corporate Firewall, Database Integration & Security Configuration.
Target Audience: PP administrators, PP developers
Biology Workshop
The first half of the workshop will feature an overview of the new Pipeline Pilot collection for Next-Gen Sequencing data. The workshop will showcase out-of-the-box capabilities for general data analysis and manipulation of NGS data, including protocols and components which map reads using standard bioinformatics algorithms, perform resequencing or de-novo assembly, and detect variants. The collection features support for standard sequencing platforms, including 454 Life Sciences, Applied Biosystems SOLiD, and Illumina (GAIIx).
Target Audience: Biologist Directors of Research and IT
Imaging and High Content Screening Workshop
The first half of the workshop will be a hands on exploration of new tools for automated Image Informatics. Emphasis on interactive learning and modeling on image data, image reporting, and dynamic image data linking to create and automate imaging workflows.
The second half of the workshop will feature an overview of the new Pipeline Pilot collections used in High Content Screening. The workshop will showcase out-of-the-box capabilities for general data analysis and manipulation of HCS data, including protocols and components for assay performance, quality control and image data linking. Sample protocols for GE Incell, Cellomics (Array Scan and Store) , Perkin Elmer (Opera and Columbus) and BD Pathway will be available.
Target Audience: Biologist, Directors of Research and IT, Image Processing Engineers, High Content Screening Scientist, Pathologist and Informaticians
Best Practices in Protocol Development
This workshop will be split into two parts (across the coffee break). Part 1 will focus on basic good practices for protocol development. Following the break, more advanced topics will be discussed. Topics may include:
- Understanding data pipelining and designing protocols to make best use of it
- Basic protocol building with Design Mode (new feature in Pipeline Pilot 8.0 – excellent opportunity to show it off)
- Merging, joining, and caching: Optimizing computationally/memory-intensive operations
- Refactoring protocols: Reviewing completed protocols to see where efficiencies can be obtained
- Help with Reporting protocols: Understanding containers and other general concepts
- Interactive reporting: Good practices when linking protocols
Target audience: The first part of the workshop is intended primarily for relatively new users of Pipeline Pilot, and for those who are interested in learning more about new Pro client features in Pipeline Pilot 8.0. The second part of the workshop will feature more advanced topics, that would particularly suit intermediate-level protocol developers.
Discovery Studio: Structure-Based Molecular Modeling
This workshop includes two hands-on tutorials in the area of structure-based molecular modelling (docking and pharmacophores) which will introduce you to the newest innovations in Discovery Studio usability and functionality
Target Audience: New Discovery Studio users seeking to do structure based calculations and Current DS users who would like to reaffirm their skills and explore best practices in structure analysis.
Automating Workflows in Materials Science
The workshop will demonstrate two different approaches for automating calculations using Materials Studio. In the first part of the workshop, MaterialsScript will be used to automate a workflow that is industrially relevant. In the second part of the workshop, the new Materials Studio Collection in Pipeline Pilot will be used to automate the same workflow.
The strategies and techniques that participants learn will be applicable to broad classes of materials including organic molecules, polymers, catalysts, and semiconductors.
Target Audience:
Current users of Accelrys Materials Studio
Enabling a Paperless Lab with an Electronic Lab Notebook (ELN)
This workshop is designed for scientists and research IT wishing to understand how an ELN can be used within the lab to replace the paper notebook. The audience will learn how an ELN can remove the tedious and time consuming process of documenting experiments with a highly efficient electronic environment, enabling scientists to rapidly document, secure, share, access and re-use experimental information within and across project groups.
Target Audience: Scientists and research IT wishing to understand how an ELN can be used within the lab to replace the paper notebook.
*Please note that titles and abstracts are subject to change.