Elke Lang

SWEET - WWW-based rapid 3D construction of oligo- and polysaccharides.

UNLABELLED SWEET is a WWW-based tool which rapidly converts the commonly used carbohydrate sequence information directly into a preliminary but reliable 3D model which can be visualised and written to files in several ways. AVAILABILITY SWEET is accessible via the Internet at http://www.dkfz-heidelberg.de/spec/. CONTACT a. bohne@dkfz-heidelberg.de or w.vonderlieth@ dkfz-heidelberg.de SUPPLEMENTARY INFORMATION The current version of SWEET generates only one conformation out of a manifold. Several authors have analysed possible conformations of high-mannose N-linked glycans using a combination of NMR methods and computational approaches showing that such molecules are rather flexible populating normally several conformations for each glycosidic linkage. The displayed model exhibits for all glycosidic linkages a conformation which is in accordance with the reported variations of Phi, psi and omega values for specific linkage (see http://www.dkfz-heidelberg. de/spec/sweet2/doc/input/sba_example.html).

LINUCS: linear notation for unique description of carbohydrate sequences.

The use of proteomics databases has become indispensable for daily work of molecular biologists, but this situation has not yet been achieved for carbohydrate applications. One obvious reason is that existing data collections are only rarely annotated and no cross-linking to other resources exists. The existence of a generally accepted linear, canonical description for carbohydrates which can be readily processed by computers will enable efficient automatic cross-linking of distributed carbohydrate data collections by serving as a unique and unambiguous database access key. Various possibilities to derive a canonical notation are discussed. They can be divided into attempts that require structure description alone and alternatives that profit from the fact that a preferred graph direction (non-reducing to reducing end) exists within the structure. To open a fruitful discussion among glycoscientists a possible solution is presented where the reducing monosaccharide unit is selected as graph root and linkage information is used to define the priority of the various branches. A Web interface (http://www.dkfz.de/spec/linucs/) has been created that directly converts the commonly used extended representation of complex carbohydrates into the preferred canonical description or into its inverted form.

SWEET-DB: an attempt to create annotated data collections for carbohydrates

Complex carbohydrates are known as mediators of complex cellular events. Concerning their structural diversity, their potential of information content is several orders of magnitude higher in a short sequence than any other biological macromolecule. SWEET-DB (http://www.dkfz.de/spec2/sweetdb/) is an attempt to use modern web techniques to annotate and/or cross-reference carbohydrate-related data collections which allow glycoscientists to find important data for compounds of interest in a compact and well-structured representation. Currently, reference data taken from three data sources can be retrieved for a given carbohydrate (sub)structure. The sources are CarbBank structures and literature references (linked to NCBI PubMed service), NMR data taken from SugaBase and 3D co-ordinates generated with SWEET-II. The main purpose of SWEET-DB is to enable an easy access to all data stored for one carbohydrate structure entering a complete sequence or parts thereof. Access to SWEET-DB contents is provided with the help of separate input spreadsheets for (sub)structures, bibliographic data, general structural data like molecular weight, NMR spectra and biological data. A detailed online tutorial is available at http://www.dkfz.de/spec2/sweetdb/nar/.

Do we need a Unique Scientist ID for publications in biomedicine

BackgroundThe PubMed database contains nearly 15 million references from more than 4,800 biomedical journals. In general, authors of scientific articles are addressed by their last name and forename initial.DiscussionIn general, names can be too common and not unique enough to be search criteria. Today, Ph.D. students, other researchers and women publish scientific work. A person may not only have one name but several names and publish under each name. A Unique Scientist ID could help to address people in peer-to-peer (P2P) networks. As a starting point, perhaps PubMed could generate and manage such a scientist ID.SummaryA Unique Scientist ID would improve knowledge management in science. Unfortunately in some of the publications, and then within the online databases, only one letter abbreviates the authors forename. A common name with only one initial could retrieve pertinent citations, but include many false drops (retrieval matching searched criteria but indisputably irrelevant).

Automatic assignment of bond orders based on the analysis of the internal coordinates of molecular structures

Abstract In order to facilitate the assignment of bond orders that are needed for the force-field calculation of 3D structures based on crystallographic data, we have developed a program which generates bond orders automatically. The algorithm is based on characteristic ranges for bond lengths, bond angles and torsion angles, and these ranges can be easily changed by the user in order to adapt the routine to special classes of compounds. Most of the assignments for covalent bonds are based exclusively on the bond length ranges. A check for planarity of ring systems is necessary in order to assign aromatic systems. Incorrect assignments may occur for delocalized and heteroaromatic systems.

Carbohydrates: Second-Class Citizens in Biomedicine and Bioinformatics?

Carbohydrates differ from the two other classes of biological macromolecules (proteins and DNA) in two important characteristics: their residues can be linked by many different linkage types and they can be highly branched molecules. Carbohydrates contain a potential of information content which is several orders of magnitude higher in a short sequence than that of any other biological macromolecule. Carbohydrates on proteins can influence their properties and be involved in many biological recognition processes. Unlike peptides and proteins, oligosaccharides cannot yet be characterized in terms of their secondary structural or three-dimensional motifs, although it may be possible to develop a descriptive system. Certain specific conformations of the carbohydrates may play a key role for the recognition process. Various conformations of the carbohydrate ligands also influence the enzyme-substrate processing pathway. In addition, the inherent flexibility of many oligosaccharides in solution leads to structures that are frequently best described as ensembles of distinct conformations. Understanding conformational behaviour of carbohydrates is of cardinal importance and represents a challenge for theoretical conformational analysis techniques, such as systematic conformational search, molecular dynamics and Monte Carlo sampling.

Automatic conformation analysis applied on large databases

Abstract To study conformational phenomena of large sets of constitutionally similar structures, the lack of a relationship between the 2D and the 3D representations as well as between substructure and complete structure representations must be remedied in order to gain access to the information stored with them. Therefore, several tools for the search and matching of different structure representations have been established. The attempt presented here considers the configurations of structures which are derived from their 3D coordinates and achieves a separation of the structures into distinct classes of stereoisomers before applying the conformation parameter calculation routines to the 3D coordinate sets. All of the user action are performed by using a graphics-orientated menu. The output files of the calculations can be universally used as input files for various visualisation and statistics tools.

PMD2HD – A web tool aligning a PubMed search results page with the local German Cancer Research Centre library collection

BackgroundWeb-based searching is the accepted contemporary mode of retrieving relevant literature, and retrieving as many full text articles as possible is a typical prerequisite for research success. In most cases only a proportion of references will be directly accessible as digital reprints through displayed links. A large number of references, however, have to be verified in library catalogues and, depending on their availability, are accessible as print holdings or by interlibrary loan request.MethodsThe problem of verifying local print holdings from an initial retrieval set of citations can be solved using Z39.50, an ANSI protocol for interactively querying library information systems. Numerous systems include Z39.50 interfaces and therefore can process Z39.50 interactive requests. However, the programmed query interaction command structure is non-intuitive and inaccessible to the average biomedical researcher. For the typical user, it is necessary to implement the protocol within a tool that hides and handles Z39.50 syntax, presenting a comfortable user interface.ResultsPMD2HD is a web tool implementing Z39.50 to provide an appropriately functional and usable interface to integrate into the typical workflow that follows an initial PubMed literature search, providing users with an immediate asset to assist in the most tedious step in literature retrieval, checking for subscription holdings against a local online catalogue.ConclusionPMD2HD can facilitate literature access considerably with respect to the time and cost of manual comparisons of search results with local catalogue holdings. The example presented in this article is related to the library system and collections of the German Cancer Research Centre. However, the PMD2HD software architecture and use of common Z39.50 protocol commands allow for transfer to a broad range of scientific libraries using Z39.50-compatible library information systems.

Matching of 2D and 3D Structure Descriptions in the Cambridge Crystallographic Data Files: Outline of the Method and some Applications

A method has been developed to join together two-dimensional and three-dimensional structure descriptions in the Cambridge Crystallographic Data Files. From the three-dimensional structure description a two-dimensional connection table is derived and then compared with the original two-dimensional connection table. Despite differences in the information which is used to describe the atoms of the two molecule graphs unambiguously, the algorithm minimizes the matching possibilities by classifying the atoms. The relation between the two graphs helps in marking structure templates in the three-dimensional images of molecules as well as making investigations on the three-dimensional structure and attaching the results to the two-dimensional graph representation.

The Integration of the Cambridge Crystallographic Data Files into the Relational Information Network of the German Cancer Research Center

The Cambridge Crystallographic Data Files (CCDF) represent the largest collection of published crystallographic data. The CCDF, originally implemented in a hierarchical data base design, have been converted to a relational architecture in order to make data access compatible with the concepts developed for the spectroscopic information system, SPEKTREN II, of the German Cancer Research Center (DKFZ), i.e.: the procedures for handling chemical structures and related user-interface software. In addition, the 3-D data of the CCDF can be stored, modified and used to generate starting conformations for Molecular Modeling calculations.