ACS Chemical Information Division (CINF)
Fall, 1997 ACS National Meeting
Las Vegas (Sept. 8 - 11)
MONDAY AM / PM
MONDAY AM / PM
TUESDAY AM / PM
WEDNESDAY AM / PM
|Skolnik Award Symposium|
|J. Gasteiger, Organizer, Presiding|
The synthesis planning in AIPHOS - Recent development.
K. Funatsu, H. Yoshino, Department of Knowledge-Based Information Engineering, Toyohashi University of Technology, Tempaku, Toyohashi 441, Japan.
For practicality of organic synthesis design system AIPHOS, the method for automatic selection of suitable starting material and reduction of strategic sites and the corresponding reaction paths, has been developed. Skeletons of both the starting material selected from library and target structure are overlapped. In a case that strategic sites already obtained by other module of AIPHOS are included in the above overlapped skeleton, the corresponding strategic sites are discarded because the sites break the starting material skeleton. It has been shown that the program prepared in this study can generate useful synthesis path keeping skeleton of the selected starting material.
Electronic information for organic synthesis - how to make the best use of it.
G. Grethe, D. Hounshell, R. W. Snyder, MDL Information Systems, Inc., 14600 Catalina Street, San Leandro, CA 94577
Over the last few years the amount of reaction information available electronically in-house or online from large databases has increased dramatically. Combined with data from smaller specialty databases, such as those for protecting groups or solid-phase synthesis, this information becomes increasingly difficult to manage by the end-user chemists. This paper will discuss the problems associated with reaction retrieval and steps taken to alleviate some of the problems. These steps include improvements in user interface to help both novice and expert users and classification of reactions and data clustering to increase the effectiveness of managing search results. Examples taken from the very recent literature will be used to demonstrate these improvements.
Browsing in reaction databases: Creating order out of chaos.
J. R. Rose Department of Computer Science, University of South Carolina, Columbia, S.C. 29206
The explosive growth in the size of reaction databases has resulted in a new set of problems. One of the most pressing problems is not how the data is stored but how the user navigates through such vast amounts of information. Query methods that were adequate for reaction databases comprising tens of thousands of reactions are woefully inadequate when the database grows by one or two orders of magnitude. An approach to browsing based on interactive hierarchical classification will be presented.
Interactive and complex visualization of chemical reaction schemes.
R. Deplanque FIZ-Chemie GmbH, Postfach 12 60 50, D-10593 Berlin, Germany
The usual method to describe chemical reactions graphically is to combine structural formulas of reactants and products in a two-dimensional static form. In the ChemInform database the total synthetic pathway is available. It is not possible to include the kinetic and thermodynamic aspects of the reaction. Important factors such as stereoselectivity cannot be shown properly. New ways of visualising chemical reactions allow interactions with the complex parameters of the total reaction within 3D hyperspace, thus increasing ones understanding of the underlying chemistry and shortening the development time of new compounds. Using this new visual approach we will discuss the advantages and the limitations of multi- and hypermedia methods in science and to which extent a combination of modern computer tools can help the scientist in the understanding of nature.
Software to aid in drug design, synthesis, and evaluation
Jorgensen Department of Chemistry, Yale University, New Haven, Connecticut 06520-8107
The CAMEO and MCPRO programs are being applied to aid drug discovery. CAMEO is used to predict products of organic reactions given the starting materials and conditions. This allows evaluation of the feasibility of proposed synthetic routes, provides SAR data, and assists in the design of optimal combinatorial libraries. CAMEO is also being used to evaluate the stability of drug candidates by analyzing the outcomes of degradative reactions including hydrolyses, thermolyses, photolyses, and oxidations. MCPRO is used at the design level to evaluate the structures and free energies of binding for protein-ligand complexes. Monte Carlo statistical mechanics is employed with both free energy perturbation and linear response approaches. Some systems that have been treated include thrombin with sulfonamide inhibitors, FKBP and FK506 mimics, and cyclophilin with cyclosporin derivatives.
The confluence of rational and combinatorial design methodologies.
S. D. Kahn Molecular Simulations, Inc., 9685 Scranton Road, San Diego, California 92121 USA
Merging rational drug design techniques that employ analog information to develop 3D structure-activity hypotheses, and using these hypotheses to focus combinatorial libraries to optimize activity, leverages the wealth of tools already developed to identify diverse chemical templates for the design of entirely new libraries optimized on a given biological response. Results will be presented that mimic multiple iterations of the drug design process to validate the methodology, and in so doing argue for the use of 3D information in the combinatorial process. Throughout the process, necessary links to corporate information (e.g., stored within MDL's ISIS databases) will be highlighted.
Computer-aided drug design: Current state and future perspectives.
H. Kubinyi, H. J. Boehm, Drug Design, BASF AG, D-67056 Ludwigshafen, Germany, and Computational and Structural Chemistry, Hoffmann-La Roche AG, CH-4070 Basle, Switzerland.
With the ongoing progress in protein crystallography and NMR, rational approaches in drug design become more and more important. Many successful lead optimizations prove the value of structure-based approaches. After the pioneering work of the Kuntz group at UCSF (program DOCK) and Howe and Moon at Upjohn (program GROW), the de novo-design program LUDI was developed at BASF. It incorporates and combines 3D searches of structural databases with growing and linking modes. In addition, it is capable to consider the flexibility of a ligand. A scoring function provides a rank order of the results. Some practical examples of the successful application of LUDI in the design of enzyme inhibitors and other protein ligands will be presented. Further developments of de novo-design programs will consider the synthetic accessibility of the proposed ligands and will build ligands within the binding site, following the principles of combinatorial chemistry. These options are important steps towards the goal of an automatic design of biologically active ligands.
A field-based approach to molecular similarity with applications to small molecules and proteins
G. M. Maggiora, J. Mestres, D. C. Rohrer, Computer-Aided Drug Discovery, Pharmacia & Upjohn, 301 Henrietta St., Kalamazoo, MI 49007
Field-based approaches to molecular similarity rely primarily upon matching the steric and electrostatic fields of the set of molecules being compared. Essentially all studies to date have been confined to small molecules. Recently, however, work in our laboratory has shown that field-based approaches can be applied to proteins as well. In fact, the results obtained thus far suggest that field-based similarity matching of proteins provides a robust procedure for aligning three-dimensional protein structures that avoids the bias introduced in many other procedures based upon sequence alignment. The presentation will summarize our approach to small-molecule field-based similarity matching, as implemented in the program MIMIC, and its application to the matching of protein structures. Examples of the three-dimensional alignment of proteins from several families such as matrix metalloproteinases will be presented.
Descriptors that outperform substructures in diversity analysis.
Y. C. Martin, R. D. Brown, E. A. Danaher, J. DeLazzer, I. Lico, Abbott Laboratories; D-47E, AP10/2; Abbott Park, IL
In our earlier evaluation of molecular descriptors for diversity analysis, we found that traditional substructure keys more successfully distinguished active from inactive molecules than did 3D descriptors generated from the distances between atoms. This report describes 3D descriptors generated from the location of hypothetical site points complementary to pharmacophoric atoms in the molecules. Classification of atoms was accomplished with hundreds of Daylight SMARTS targets to describe Hbond donors & acceptors, positively & negatively charged atoms, and hydrophobic centers. The location of the site points was determined from crystallographic packing or 6-31G** calculations. These new descriptors more successfully distinguish active from inactive molecules in our most diverse dataset. Additionally, using them in a clustering analysis results in larger bioequivalent clusters compared to the best 2D descriptors.
Pharmacophores in Drug Discovery.
G. W. A. Milne, S. Wang, M. C. Nicklaus, Laboratory of Medicinal Chemistry, NCI, NIH, Bethesda, MD 20892.
Chemical bind to enzymes with a variety of forces such as hydrogen bonds and the atoms in a substrate which bind to the enzyme constitute the pharmacophore - that part of the molecule which is responsible for the biological activity. Any molecule which contains that pharmacophore is capable in principle of binding to the enzyme. A compound that can bind in the active site of an enzyme can behave as an inhibitor because it can compete for that site with the enzyme's normal substrate. If the reaction that is inhibited is essential to a disease process, the inhibitor is a potential lead drug. This has led to the technique of pharmacophore searching in which the NCI database is examined for compounds which contain a pharmacophore - defined constitutionally and geometrically. Such compounds can be retrieved from the NCI repository and bioassayed. The proportion of the compounds retrieved by pharmacophore searches which are biologically active ranges from 10% to 50%. The method has been applied successfully to the discovery of inhibitors of HIV protease and integrase, of activators of protein kinase C and a variety of other bioactive compounds.
Molecular similarity using two-dimensional representations of structures.
W. G. Richards, D. D. Robinson, Physical and Theoretical Chemistry Laboratory, Oxford University, South Parks Road, Oxford OX1 3QZ, UK.
The introduction of high throughput synthesis and combinatorial chemistry has necessitated a huge acceleration of the time required to compute molecular similarities. One way in which this may be achieved is to work in two dimensions. At the same time the details of three-dimensional shape are so important that they must be incorporated. Representations produced by non-linear mapping techniques have the required properties. As with three-dimensional similarity, the molecules to be compared must be superimposed and optimally aligned. Techniques such as invariant moments permit this and also offer novel ways in which similarity may be computed. The procedures are based on techniques originally developed for optical character recognition and offer similar possibilities for speed.
3D Structures and the bioinformatics revolution
D. H. Smith MDL Information Systems, Inc., 14600 Catalina St., San Leandro, CA 94577
Research in genomics is now a strategic component of life sciences research and discovery. This revolution requires us to examine the role of three-dimensional structures and their possible receptors with a new set of objectives. I will discuss the emerging role of structural bioinformatics in the process of target selection, and the attendant requirements for high quality structural information on candidates screened in the virtual laboratory.
Challenges and progress in structure-based ligand design.
A. P. Johnson University of Leeds, Leeds, LS2 9JT, UK
The SPROUT program provides a set of tools designed to aid the synthesis of synthetically accessible ligands which either fit a pharmacophore hypothesis or are predicted to bind strongly to a particular 3-D structure of a protein. The HIPPO module identifies potential target sites where suitable ligand atoms might be positioned for binding to the protein by H-bonding, covalent interactions, metal ligand interactions and hydrophobic interactions. Other modules dock small fragments to these sites and connect them together by an exhaustive growing process. The ALLIGATOR module provides clustering and ranking tools which permit the user to navigate through the answer set, including estimation of structural complexity as well as an empirical estimate of binding affinity. In the final step, the CAESA module is used to provide an estimate of the ease of synthesis of each of the suggested molecules. Examples of the application of SPROUT to inhibitor design will be presented.
Second generation de novo drug design methodology.
W. T. Wipke. Brian Goldman, Michael Kappler, Brett Kislin, John Lawton, Jim Arnold, Molecular Engineering Laboratory, Dept. of Chemistry, UCSC, Santa Cruz, CA 95064 firstname.lastname@example.org
The first generation of structure-based de novo design of drug candidate molecules was based on a pharmacophore model, lead compound, and/or receptor three-dimensional structure. We have developed an automated design system, INVENTON, that incorporates these design approaches. Our recent work has been bringing INVENTON to what we call the second generation level of design concepts. These are concepts that are difficult for chemists, e.g., protein and inhibitor flexibility, potential mutations that may occur in the protein, alternative binding modes, and water molecules in the active site. We hypothesize that appropriate consideration of these factors will lead to superior designs for drug candidates.
De novo screening - A search method for the discovery of novel ligands.
P. W. Rose, T. J. Marrone, Agouron Pharmaceuticals, Inc., 3301 North Torrey Pines Court, La Jolla, California 92037.
De novo screening is the process of computationally discovering fragments that specifically interact with biological targets whose three-dimensional structure is known. Fragments used in this process are either generated computationally, obtained by fragmenting existing molecules, or are small molecules that are commercially available. The fragments are docked into the target protein and the binding free energy is estimated by an empirical scoring function derived from the analysis of a large number of protein-ligand complexes. The top scoring fragments can be elaborated into ligands or they can serve as core structures in combinatorial libraries.
Years living with molecular topology.
J-E. Dubois, ITODYS, 1 rue Guy de la Brosse, 75005 Paris, France.
From the intrusion of computing in daily theoretical chemistry and information retrieval in 1965 to the present situation where micro computing and Internet modify the working habits of the chemist, many drastic changes have occurred in the chemist's activities. The classical vision of "constitutive fragments" of a structure faced a strong competition of different captures of information with the "molecular topological paradigm". Structural topology builds a graph vision of a structure with its matrix description. As a structural paradigm it leads to new languages and a global vision of environmental aspects of a site in a structure. At first it seems to oppose the "molecular fragmentation paradigm" but if computing leads to excellent topological tools it also enables a combination of the two paradigms as often complementary sources of information and modeling tools for CAD design. These changes have played a major role professionally and institutionally in various ways and at different levels.
|C. Gragg, Organizer, Presiding|
Searching the Beilstein bibliographic file on the internet: the NetFire system.
J. L. Wisniewski Beilstein Informationssysteme GmbH, Varrentrappstr. 40-42, Frankfurt/Main, Germany.
A new literature-based database and search system, called NetFire, has recently been made available on the Internet by Beilstein Information Systems. NetFire is a powerful new way for scientists in industry, government and academia to access titles, abstracts and authors of reports published in over 140 top journals in organic and medicinal chemistry. Specifically designed to take advantage of features found in the most popular web browsers available today, NetFire provides users with a tool for searching and displaying literature chemical information. The easy-to-use query-by-form mechanism permits the formulation of a great variety of inquiries. One may search by author, and by words included in the abstract or title. A search can also be restricted to a specific journal or time range. The paper will discuss in detail the design, implementation, and performance of this innovative resource from Beilstein.
Reengineering a corporate chemical information system using the latest client/server technologies: A chemist/end-user approach
P. Perraudin Questel-Orbit, Le Capitole 55, avenue des Champs Pierreux, 92029 Nanterre Cedex - France.
Chemical and pharmaceutical companies can no longer consider their in-house DBMSs for chemical structures and reactions as isolated, standalone, independent systems. Chemists need access to external chemical and other information, too, without interrupting their primary jobs. Working with five major European chemical and pharmaceutical companies, we provided their chemists a very intuitive tool, part of the development of future ChemPath Structure and Reaction System, by specifying, testing and validating the new client/server user-interface they expected. This also provided custom links to the rest of the information management system in their respective R&D centers. This paper will show how each succeeded in adapting ChemPath to its own needs and how they see the future of their global information system and end-user Windows/intranet interface.
The Derwent crop protection file, a valuable underutilized source of pesticide information.
J. D. Myers, M. D. Bauer, 1725 Duke Street, Alexandria, VA, 22314
In a world becoming more and more concerned with air, water and soil pollution, information on effective but environmentally friendly pesticides becomes more important. This paper shows how valuable information from published literature, scientific meetings and patents on pesticides can be obtained from the Derwent Crop Protection File (DCPF). Examples of searches on biological control techniques available for use in green-house grown ornamental plants, effects of certain popular pesticides on vertebrates, and novel pesticides currently under development are included. We will also show which companies are the major players in development of new pesticides. Statistical analysis on search results will recognize companies and scientists that are major contributors.
The Merck Index: Assessing usage patterns and impact on scientific communication through citation analysis.
S. Budavari. The Merck Index, Merck Research Laboratories, Rahway, New Jersey 07065
The Merck Index has been highly cited over the years and on numerous occasions has even been referred to as the "chemist's bible". This talk will present a brief overview of the results of a recent investigation of citations of The Merck Index and what they suggest with respect to how the publication is being used by scientists. In addition, inferences as to the role of the "Index" in the realm of scientific literature will be discussed, as well as the implications for future editorial policy.
SLIMS: A spectral laboratory information management system based on web technology.
A. J. Williams, A. Petrauskas, P. Jurgutis, D. Ross, V. Kulkov, Advanced Chemistry Development, Inc., 133 Richmond Street, Suite 605, Toronto, Ontario, CANADA M5H 2L3.
At ACD we have designed and implemented a system to allow corporate-wide access to analytical information, focussing specifically on spectral information. A number of efforts have been made over the years to implement flexible Laboratory Information Management Systems, LIMS. These include: filing systems for data, paper trails and log books, simple spreadsheets containing sample identifiers and information as well as software based vendor-supplied LIMS system. Software LIMS systems have failed to address the flexibility of interface and features required in an analytical environment that requires access to molecular structures and graphics intensive spectral displays. As a result of collaborative efforts with Eastman-Kodak company we have developed a web-based LIMS systems for managing spectral and associated molecular structure information. This user-friendly system links a unique sample identifier to sample information, a chemical structure or structures, associated spectra and finale reports of analysis.
Internet access to genetic sequence databases.
D. R. Jourdan, Kraft Foods, 801 Waukegan Rd., Glenview, IL 60025. Djourdan@kraft.com
In the last decade the number of genetic sequences being submitted to various databases has dramatically increased. The tools used to access these sequences have recently grown much more powerful. Email autoresponders and web databases allow information professionals and researchers to quickly locate sequences of interest based on fields such as author, organism, or sequence accession number. These records may also provide substantial structural, bibliographic, or taxonomic information. Additionally, tools such as BLAST (Basic Local Alignment Search Tool) allow for the identification of biologically similar sequences.
Comparing diversity selection methods against random selection.
T. Hurst, Tripos, Inc. 1699 South Hanley Road, St. Louis, MO 63144.
Many methods for selection of diverse and representative sets of structures from large collections have been developed over the last few years. These methods vary greatly. Some involve intense statistical methods for clustering, others simple 2D similarity comparisons, and others include 3D information such as pharmacophore triplets. In all of these methods, it is presumed that a diverse subset will increase the efficiency of biological screening. In this presentation, we will answer the pervasive question of this field: "How do I know that screening this selection is better than random screening?"
Secondary research based on the technical literature: The time factor.
M. P. Bigwood, International Technology Information, PO Box 58, Oreland, PA 19075-0058.
Content analysis of the technical literature has been used as the foundation of many technology assessments. Bibliometric analysis methods, on the other hand, are very powerful tools in areas such as strategic technology management and technical competitive intelligence. The argument has been made, however, that these methods are of little value because of the long delays that separate the time a significant technology development takes place in the laboratory and the time it appears in the published literature. In this presentation we want to share data we have collected that provides a quantitative estimate of the time required for a technical development to reach the published domain and compare that timing with a few examples of technology cycle times. As fast as today's technology might be moving, by historical standards, we will see that, for specific cases we looked at, analysis of the published literature in general, and of the patent literature in particular, is still a source of valuable information.