Synthesis design

The system WODCA (Workbench for the Organization of Data for Chemical Applications) has been developed over the last 10 years to assist in the planning of the synthesis of individual target compounds [1,2]. Work of recent years now also enables its use in designing entire libraries of compounds.

As the acronym implies, WODCA provides a workbench and a series of tools for designing a synthesis. Two of the more important tools are search methods for defining strategic bonds and methods for searching for similar ioo [1] S

Figure 1. Example of a target structure (TIBO). Strategic bonds (in this example only carbon-heteroatom bonds) are rated by WODCA (relative to the bond with the highest rating 100). The numbers in brackets are the ranking ofthe bonds.

compounds in libraries of available starting materials [3]. Strategic bonds comprise bonds where a target should be broken apart (disconnected) to obtain synthesis precursors which would react back to the target structure through efficient organic reactions.

For each bond the two possible heterolytic bond breakings are explored and evaluated by physicochemical effects such as bond polarity [4], and the stabilization of charges by inductive [5], resonance [6], and polarizability effects [7]. These methods are calculated by rapid procedures collected in PETRA (Parameter Estimation for the Treatment of Reactivity Applications) [8,9]. Furthermore structural effects such as branching of bonds, positions at ring systems, and stereochemical centers are taken into account. The rating of the strategic bonds of a molecule is then scaled to a value between 0 and 100. Figure 1 shows the results obtained for the compound TIBO [10].

The searches for available starting materials are performed on the basis of similarity criteria that are either based on structural similarity or on generalized reactions [3].

In order that WODCA can be used for the design of the synthesis of libraries of compounds, substructure search methods have been included in the system. These, in conjunction with the search for strategic bonds, can provide a series of representatives of certain classes of compounds that can act as precursors for the synthesis of an entire library.

We will briefly explain how these methods in WODCA can be utilized in a sequence of steps to derive a series of starting materials for a combinatorial library.

1. A typical representative of a desired library is input.

2. A search for strategic bonds is performed. This points out the bonds that should be retrosynthetically broken.

3. One or more strategic bonds are broken. In the decision which bonds should be broken, the chemist/user can follow the rating suggested by the program, or she/he can override it and select her/his own strategic bonds.

4. The strategic bonds will be broken and the free valencies thus obtained will be appended with atoms or groups of atoms in a way that takes care of the charges generated after heterolysis.

5. These synthesis precursors are presented to the user in order that she/he removes atoms or groups at those positions where the precursors should be generalized, where variations in the groups should be made in the combinatorial synthesis.

6. With these fragments substructure searches are performed in a catalog of available starting materials. This provides all members of a class of compounds contained in a catalog of chemical suppliers. Various catalogs are incorporated in the WODCA system. Mechanisms are available to load other catalogs such as in-house or corporate databases.

These steps are illustrated in Figure 2 with the example of developing a synthesis for substituted pyrazoles and are further commented:

1. 1-Pheny1-3,5-dimethylpyrazole is chosen and input as a typical representative of a desired library of compounds.

2. The rating of strategic bonds is shown in Figure 2. This rating defines the sequence in which the bonds would automatically be broken: First, the bond with the higher rating, then the bond with the lower rating.

3. The user can follow the sequence of strategic bonds suggested by the program or can change the order. Furthermore, it is possible to require several bonds to be broken automatically. In the given example, the two bonds suggested by the program were accepted as strategic to be broken.

4. In breaking these strategic bonds WODCA realizes, on the basis of calculations of physicochemical effects by methods contained in the PETRA package, that these bonds should be broken heterolytically and it determines where the charges should go. Based on these charges, decisions are made which atoms or groups, in our case hydrogen and oxygen atoms and a hydroxy group, should be added to these free sites.

5. The user has decided to remove the two methyl groups from the precursor 2-hydroxy-pent-2-ene-4-on. Various options are available to define which atoms are allowed at these open positions for the substructure search. In our case, no restrictions on the types of atoms allowed at these free sites were imposed. Furthermore, the option that allows tautomerization was invoked. For the other precursor, phenylhydrazine, all hydrogen atoms

Figure 2. Steps in the synthesis planning of a pyrazole library with WODCA (see also text): 1. given target structure (1); 2. strategic bonds rated by WODCA; 3. both bonds are broken by user; 4. precursors (2) and (3) generated by WODCA; 5. definition of open sites at these precursors at all positions where the precursors could be generalized; 6. substructure search in Fluka catalog.

144 compounds

Figure 2. Steps in the synthesis planning of a pyrazole library with WODCA (see also text): 1. given target structure (1); 2. strategic bonds rated by WODCA; 3. both bonds are broken by user; 4. precursors (2) and (3) generated by WODCA; 5. definition of open sites at these precursors at all positions where the precursors could be generalized; 6. substructure search in Fluka catalog.

on the phenyl ring were removed, basically asking for a search for any substituted phenylhydrazine.

6. The two substructure searches in the Fluka catalog contained in WODCA provided 144 hits for the 1,3-dicarbonyl-substructure (54 hits in the Acros catalog) and 23 substituted phenylhydrazines (38 phenylhydrazines in the Acros catalog). The hits obtained with the 1,3-dicarbonyl subunit contained quite a few /p-ketoesters, P-ketoamides, and substituted malonic

Figure 3. More specific substructures defined for the precursors (2) and (3). The number of hits in the Fluka catalog (and, in parentheses, in the Acros catalog) for each substructure is given.

acids. In order to exclude these, more specific substructure searches have to be performed. On the other hand, if also aliphatic hydrazines should be included, a more general substructure has to be specified. Figure 3 shows the results obtained with these variations in the definition of the query substructures.

This selection of starting materials can be entered into a combinatorial chemistry experiment or a parallel synthesis. For modeling and evaluating such a synthesis of a library these precursors can be handed over to the EROS system (see next section).

0 0

Post a comment