Reaction substructure

The next abstraction level of reaction retrieval is a so-called reaction substructure search in which both query structures arc considered as substructures. In the case of a reaction substructure search, no hydrogen atoms arc added internally during the execution of the search. Atoms which have their valencies not completely saturated are considered as open sites, where any hind ofelement could be bonded. 

Figure 5-25. First record of the hit list of the reaction substructure search (MDC number RXCI91107130).

To obtain the required product yield, the reaction substructure search is combined with a textual search using the queiy builder" of the MDL" ISIS program. 

The next example shows how different search queries can be combined to shed more light onto a series of related reactions. A reaction substructure search for reactions that break a P-O bond provided 304 reactions as hits. Figure 10.3-27 shows one of the reactions in this hit list. 

The breaking of a strategic bond and the generation of synthesis precursors defines a synthesis reaction. In the simplest case, the reaction is already known from literature. In most cases, however, the rcaaion step obtained has to be generalised in order to find any similar and successfully performed reactions with a similar substituent pattern or with a similar rearrangement of bonds. One way of generalizing a reaction is to identify the reaction center and the reaction substructure of the reaction. This defines a reaction type. 

In a reaction, bonds are broken and made. In some cases free electrons are shifted also. The rcaciion center contains all the bond.s being broken or made during the reaction as well as all the electron rearrangement processes. The reaction uhstme-ture is the structural subunit of atoms and bonds around the reaction center that has to be present in a compound in order for the reaction to proceed in the foi"ward (synthesis) direction (Figure 10,3-32). Both characteristics of a reaction can be used to. search for reactions with an identical reaction center and reaction substructure but with different structural units beyond the reaction substructure. For example, this can be achieved by searching in a reaction database. 

Figure 10.1-32. Reaction center and reaction substructure the parts of the structures with a darker gray background are the reaction center, and those with a lighter gray background are the reaction substructures which must be present to achieve the reaction In the forward direction (in this case, Michael addition).

Figure 10.3-42. Deriving the reaction substructure, A diFferent number of bond spheres around the strategic bond can be Included in the reaction substructure, thereby influencing the specificity of a search in a reaction database.

The position of the ehosen strategic bond locates the reaction center. To derive the reaction siibstrncture, the user can select the number of bond, spheres around the strategic bond which should be included. The reaction substructure obtained is then n.scd as the query for a reaction substructure search in the database. Figure 10,3-42 illustrates the first and second bond spheres around a selected strategic bond of a retrosynthetic step. 

The number of bond spheres chosen influences the specificity of tlie reaction substructure query and the result of the search. In Figure 10.3-43 the reaction substructure queiy including the first bond sphere of the retrosynthetic step of Figure 10,3-42 is shown. 

Both precursors can be used as reactants in an aldol condensation. It has to be emphasized that the chlorine atom in 4 has to be considered as a representative for any electron-withdrawing group in particular, in the case presented here, it would best be taken as an OEt group. In order to verify this proposal, a reaction substructure search is initiated in the Chcmlnform reaction database of 1997. 

Figure 10.3-54 illustrates how the reaction center is derived from the disconnection of the strategic bond and which additional bond spheres are considered in the definition of the reaction substructure search queiy. 

One of the hits found in the Chem Inform reaction database is shown in the window for reaction substructure searches in Figure 10.3-55. It fits the synthesis problem perfectly, since in the synthesis direction it forms the coumarin ring system directly, in one step. 

Figure 10.3-54. Derivation of a reaction substructure search query. The a" for atom 1 and 8 indicates that these atoms have to be aromatic in the found reactions.

Chen, L., Nourse, J. G., Christie, B. D., Leland, B. A., and Grier, D. L. (2002) Over 20 years of reaction access systems from MDL a novel reaction substructure search algorithm. J. Chem. Inf. Comput. Sci. 42, 1296-1310. 

Figure 9.15. Reaction substructure search query and some example hits. If no reacting center or mapping information is used, all three hits are found. If reading bond information is used, hit c is excluded. If both reacting atom and reacting bond information is included, then false hits b and c are excluded.

After the disconnection strategy is defined, the systems indicate the strategic bond together with their ranks. The user can now analyze the precursor or can verify the disconnection by performing a reaction substructure search in any of the interfaced reaction databases. To perform a search in the reaction database, the user can define the bond sphere to be considered as identity criterion. The first sphere, for instance, includes bonds attached to the atoms of the strategic bond. A hit is presented as a reaction with additional information from the reaction database, such as reaction condition, yield, and references. 

Of the graphical search types listed, one of the most important is Reaction Substructure Search, or RSS. The reaction indexing group at Molecular Design initiated a new project to increase the flexibility and specificity of RSS searching. That project is the subject of this paper. 

In using a substructure approach, we determined that only 58% of the 138 citing papers would have been retrieved. Thus we observe that no single approach, reaction, substructure, cited reference, or keyword, would allow for complete retrieval of related papers and therefore a comprehensive search strategy must include multiple approaches. 

Figure 10 shows the results of searches with several reaction substructure queries apphed to the Theilheimer database, on a DEC Microvax 2000. For each query, three t)q)es of searches were performed ... 

We found that the number of answers to reaction substructure queries over first-order implicit schemes was disappointing in general, compared to the amount of computer resources required to execute the search. Although many chemists will be willing to wait for those results, the authors are not conviced that a majority of users would consider the added results to be worth the decrease in performance. 

Reaction data such as yield, conditions (e.g., temperature, pressure, atmosphere, pH), and additional comments like enantiomeric excess and photochemistry are potentially important to narrow down the results of reaction substructure... 

Substructure searching of explicit reaction schemes has been implemented in the CASREACT system by enumerating all the individual reactions and indexing them individually. The application of reaction substructure searching techniques to query implicit reaction schemes (where the product of one reaction is the reactant of another in the database) and analogous reaction schemes (where the reaction center of a reactant in one reaction is present in the product of another) has also been explored. ... 

---