CHIRBASE

CHIRBASE Database Current Status and Derived Research Applications using Molecular Similarity, Decision Tree and 3D Enantiophore Search... 

The need for an overall and combined chemical structure and data search system became clear to us some time ago, and resulted in the decision to build CHIRBASE, a molecular-oriented factual database. The concept utilized in this database approach is related to the importance of molecular interactions in chiral recognition mechanisms. Solely a chemical information system permits the recognition of the molecular key fingerprints given by the new compound among thousands of fingerprints of known compounds available in a database. 

In this chapter, we will discuss the present status of CHIRBASE and describe the various ways in which two (2D) or three-dimensional (3D) chemical structure queries can be built and submitted to the searching system. In particular, the ability of this information system to locate and display neighboring compounds in which specified molecular fragments or partial structures are attached is one of the most important features because this is precisely the type of query that chemists are inclined to express and interpret the answers. Another aspect of the project has been concerned with the interdisciplinary use of CHIRBASE. We have attempted to produce a series of interactive tools that are designed to help the specialists or novices from different fields who have no particular expertise in chiral chromatography or in searching a chemical database. 

As we have already pointed out, the first objectives of CHIRBASE project are to ... 

Some details of the database are available on the Web [6] and consist today of about 61 000 different chiral separations (Table 4-1). Between 12 000 and 15 000 new separations are stored continuously each year in CHIRBASE. 

ISIS databases are hierarchical, so CHIRBASE was designed to incorporate about 60 data fields on several levels of detail (the main fields are listed in Table 4-2). The first level contains the molecular structure of the sample combined to the molecular structure of the CSP, producing a unique location or entry for a specific sample-CSP couple. Consequently, in the current version of CHIRBASE, which contains 40 000 entries, one entry corresponds to the separation of one sample on one CSP and contains in different sublevels a compilation of all the references and the various analytical conditions available for this separation. 

If some fields may be empty in the sublevels, all the fields in the main level are required for each entry. A new chiral separation record can be added in CHIRBASE solely if the authors correctly identify both sample and CSP. Since the beginning of the project, our policy has been to contact the authors of all publications containing incomplete, ambiguous or inconsistent data and to ask for additional information. Providing the separations with unique case numbers helps us considerably in this essential task, and also facilitates avoiding redundancies in the database. When chiral separations are reported for the second time in a new publication with exactly the same chromatographic conditions, this is stated in a footnote added in the field comments . In this field, miscellaneous information that cannot appear elsewhere are listed (detection limit, description of a reported chromatogram, racemization study, mobile phase limitations, etc.). 

The most complicated aspect of the CHIRBASE project effort is the actual incorporation and validation of data. It is largely due to the complexity of the problem and to the difficulty of extracting and interpreting the relevant information, since the vast majority of all useful data is disseminated in the papers rather than in a user-readable or a computer-readable form. 

Fig. 4-3. Molecular similarity searching of AZT in CHIRBASE. (Compounds reported in Refs. [7-12].)...

In addition to addressing the practicalities of database maintenance, we also aim to enhance the range of the facilities available, to make the information within CHIRBASE more readily accessible to users. CHIRBASE contains two form-based applications for query building designed to help novice or expert users to formulate queries the query menu and the automatic search tool. 

CHIRBASE provides integrated responses from single questions, as well as from combinatorial questions constructed on the basis of any specific query corresponding to one or several field(s) occurring in the database. With the molecular structure of a sample in hand, the search can be conducted interactively from the query menu form. 

To address these issues regarding adequate structural searching in CHIRBASE, some facilities have been recently added to the user interface. The result is the automatic generation and search of strategic 2D query structures defined with the help of the following commands (Fig. 4-5) ... 

Auto Search This button initiates from a structure query two or three automated series of search exact and substructure searches in local desktop versions exact, substructure and similarity searches in network version (under ISIS/Host). All the result lists are saved in CHIRBASE using exact-auto , SSS-auto and SIMXX %-auto names. XX is the highest similarity search value (from 80 % to 40 %) allowing to retrieve hits in CHIRBASE. The records in all the lists are unique. The SSS-auto list does not include records that are in the exact-auto list. The SIMXX %-auto list does not include records that are in exact and SSS-auto lists. 

Table 4-3. Search results from CHIRBASE 3D with enantiophore queries built from the Whelk CSP.

From these preliminary results, we can assume that such enantiophore queries could be used to search in a database of compounds with unknown enantioselectiv-ities. With respect to the percentage of resolved samples that are retrieved in CHIRBASE, the resulting list should contain a similar yield of compounds providing favorable specific interactions with the CSP receptor binding sites. 

Starting from a collection of samples remarkably well resolved (alpha > 6) on Chiralcel OD (Cellulose tris(3,5-dimethylphenylcarbamate) coated on aminopropyl silica), a putative three-point enantiophore for binding to CSR was derived (Fig. 4-10). This enantiophore query was used to search (CFS 3D search) within a list comprising 4203 compounds tested on Chiralcel OD. From this search domain of CHIRBASE 3D, 191 structures were found to match the enantiophore. 

Designed from CHIRBASE-3D, CHIRSOURCE provides 30 000 structures in terms of configurational diversity, most of them easily available by semipreparative scale on corporate installation or in dedicated companies with minor further optimization. 

The similarity matrices are constructed by one in-house program developed inside CHIRBASE using the application development kit of ISIS. They contain the similarity coefficients as expressed by the Tanimoto method. In ISIS, the Tanimoto coefficients are calculated from a set of binary descriptors or molecular keys coding the structural fragments of the molecules. 

The purpose of this study is only intended to illustrate and evaluate the decision tree approach for CSP prediction using as attributes the 166 molecular keys publicly available in ISIS. This assay was carried out a CHIRBASE file of 3000 molecular structures corresponding to a list of samples resolved with an a value superior to 1.8. For each solute, we have picked in CHIRBASE the traded CSP providing the highest enantioselectivity. This procedure leads to a total selection of 18 CSPs commercially available under the following names Chiralpak AD [28], Chiral-AGP [40], Chiralpak AS [28], Resolvosil BSA-7 [41], Chiral-CBH [40], CTA-I (microcrystalline cellulose triacetate) [42], Chirobiotic T [43], Crownpak CR(-i-) [28], Cyclobond I [43], DNB-Leucine covalent [29], DNB-Phenylglycine covalent [29], Chiralcel OB [28], Chiralcel OD [28], Chiralcel OJ [28], Chiralpak OT(-i-) [28], Ultron-ES-OVM [44], Whelk-0 1 [29], (/ ,/ )-(3-Gem 1 [29]. 

Since this current study is restricted to the best enantioseparations (a > 1.8), it is quite clear that the tree does not accurately reflect the full information contained in CHIRBASE. 

---