Database Reaction databases, retrieval systems

This tutorial describes briefly some of the search capabilities possible with Cheminform RX and MDL" ISIS used as the retrieval system. In this tutorial, the CIRX database. of the years 1992-1996 arc used, containing altogether 334 855 reactions. 

The reaction database compiled on Biochemical Pathways can be accessed on the web and can be investigated with the retrieval system C ROL (Compound Access and Retrieval On Line) [211 that provides a variety of powerful search techniques. The Biochemical Pathways database is split into a database of chemical structures and a database of chemical reactions that can be searched independently but which have been provided with efficient crosslinks between these two databases. 

And last not least, we will have to see further improvements in the graphical user interfaces of software systems and the retrieval systems of databases in order to make software and databases more acceptable to the chemical community at large. Software and databases should speak the language a chemist is used to, with hand-drawn chemical structures and reaction equations, or even imderstand the spoken word - and only provide the desired information selectively, not buried in a phe of unnecessary output. 

The next step up in integration places as much information as possible in the chemical structure or reaction DBMS. For example, Molecular Design Limited s MACCS system has been used to store structures, reactions, biology, spectra, inventory, and document index information. This approach provides a good degree of data integration but suffers from poor performance for large databases. Additionally, the retrieval capabilities of the structure DBMS are inadequate for certain types of information. 

Reaction retrieval systems are usually built without due connection with systems built earlier for the retrieval of compounds. This paper discusses options for a file organisation for a unified structural database of compounds and their reactions. Such a database has a potential for improved retrieval capabilities and high browsability. 

The 3-centre reactions are all known (Figure 4). They are equivalent to the 3-centre carbene reactions. It is important to note at this point that the program does not contain a reaction database which could be used to retrieve information. The only system inherent information is the given R-matrix and the valency of carbon. Thus the complete set of 3-centre carbene reactions was generated without chemical knowledge, but by the solution of the combinatorial problem. 

Besides the data derived from the schemes, the metafiles (the pure graphic) and, of course, the data from the textual input (title, abstract, bibliographic data) will also be stored in the base file. This will be the source for indexes or subfiles, and for reaction databases for existing retrieval systems. As far as the used retrieval system will allow, all of these data except the display-only graphic metafile may be searchable. 

It is planned to produce an online version of the database, which will be implemented on STN International. For this purpose, a data file in standard format that can be used under Messenger software will be derived from the base file. (Stereochemical information will be kept in this file, even though Messenger cannot handle this information at present). The enhancements to Messenger developed at CAS for CASREACT, the online reaction retrieval system developed by CAS, will also be used for this database. The display capability of Messenger will, in this configuration, also allow for the display of the whole (multi-step) reaction schemes. 

On the other hand, the data in the base file may be converted into an SMD file (Standardised Molecular Data) which is used by the companies of the CASP pool in Europe for interchanging substance and reaction data. Interfaces to commercially available reaction retrieval systems, using the SMD data structure, exist or are being developed. In this way, in-house versions of the database to be used with REACCS, ORAC, SYNLIB, etc., will be made available. 

For purposes of comparison, the numbers of articles related to the most representative recent synthetic organic reactions, i.e., chiral reactions [24, 28, 31, 57-67], metathesis reactions [68-79], and cross-coupling reactions [24, 29, 30, 80-90], are shown. Information retrieval conducted with the Chemical Abstract Database SciFinder System on January 26, 2012 (total data 31,409,203) produced the following results for the three types of reaction ... 

Similarity searching in reaction databases has gained widespread interest since it became available (in REACCS and ORAC) about two years ago. The definition of similarity of reactions in these reaction retrieval systems is based on common substructural features in the reaction centres and in the reactants. Without doubt, these definitions of reaction similarity are of great usefulness. However, we wanted to broaden the discussion of reaction similarity and search for additional helpful ideas. To achieve this we took a totally different approach to specify the similarity of reactions. Our approach is based on the idea that chemists often want to know the exact conditions for running a reaction. We assumed that similar reactions should be those that proceed under similar reaction conditions. 

In this paper we describe how the ORAC reaction database system has been integrated with the LHASA synthesis planning program. This interface allows literature precedents for synthetic steps proposed by LHASA to be retrieved from ORAC and displayed to the user. The problem of sequence searching in reaction database systems is also explored and approaches to this problem are described. Finally, the potential use of synthesis planning systems to identify and classify links between reactions in ORAC for use during sequence searching is considered. 

The primary purpose of a reaction database system is to provide quick and easy access to the reaction literature. Users of synthesis planning programs have a need for such access to the literature in elaborating routes suggested by an analysis. An interface between LHASA and ORAC has recently been developed to provide such a link. The interface allows literature precedents for synthetic steps proposed by LHASA to be retrieved automatically from ORAC s database of over 120,000 reactions and to be displayed to the user on an ORAC Display Form. Company databases can also be accessed to search and display in-house preparative methods and processes in conjunction with databases supplied by ORAC Ltd. 

In the following sections, the database acronym/name is given followed by the producer, availability (retrieval systems or hosts), time coverage, and update frequency in parentheses. To enhance the short descriptions, the same reaction example from a publication by Goldsmith and Soria with a total of 11 reactions is used when available to illustrate the differences and similarities between the databases (see Section 4.3). 

Most in-house systems permit links to other database management software, in particular to relational database systems such as ORACLE, in order to integrate large amounts of textual or numeric data that already exist there with reaction databases. As many of the reaction retrieval systems mentioned here are able to import and export files in MolFile or RxnFile formats (originally proprietary file formats of MDL, later published), or in the SMD format, even exchange of reaction databases between different retrieval systems is often possible. ... 

SYNLIB is a very special reaction retrieval system, which differs significantly from other systems in terms of the underlying philosophy and functionality. It has a built-in fuzziness, such that usually not the entire (sub)structures entered are matched in the database, but only those parts of the query that the user declared as target (Figure 10). Target centers are usually reaction centers plus some parts of the environment . " ... 

ORAC was developed by A. P. Johnson et al., at Leeds University, and was later produced by ORAC Ltd. Since the takeover by MDL, ORAC has no longer been upgraded, but its core database (about 65000 reactions, coverage mainly since 1980, but with a significant number of earlier reactions) survives as part of MDL s RefLib (.see Section 3.2.3). ORAC had a few outstanding features that appeared only later in other reaction-retrieval systems, or are still missing in them thesaurus-based reaction classification (see Section 4.11), both bonds or atoms definable as reaction centers, both explicit... 

While for a long time the in-house reaction-retrieval system market was dominated by REACCS and ISIS, with almost every reaction database being available in the REACCS/ISIS format, there are now a number of competitors. Some of the Synopsys reaction databases can be run on Synopsys Accord, a structure- and reaction-handling add-on for Microsoft Access and Excel. Chem-X, a modular software system from Chemical Design Ltd. includes (since April 1993) the reaction... 

---