Markush structure can be generated automatically from a molecule library using R-group decomposition or Composer. Compared with R-group decomposition, Composer capable of calculating the optimal scaffold automatically and can generate not only combinatorial but also complex (patent like) Markush structures. Composer algorithm can be the optimal solution to automatically create virtual combinatorial libraries based on existing compound sets. These libraries covering not only the original starting structures but all possible structural combinations of them. This way, e.g. by enumerating these covered structures you can easily find new potentially effective compound. Or you can use the created Markush structure as a starting point to create your patent claims.
Markush Composer capable of generating Markush structures based on the list of molecules. MCS based algorithm recognize the common parts of the input molecules and further heuristics try to optimize the size, orientation and R-group attachments of the generated scaffold to find the best Markush core fitting the preset user preferences. If the input data set can't represent with single Markush structure after a simple MCS based clustering multiple Markush scaffold generated. The built-in heuristics try to minimize the number of created Markush structures, and the complexity (R-groups number, fragments number, and Library space) of individual Markush structures. Depending on the user settings in a further iterative process the algorithm can find the common parts of the first lever R-group fragments and capable of generating nested R-groups. Optionally the user can define a scaffold and this case the algorithm try to use the preset scaffold as the starting point instead of the default MCS based calculation.
Compared to simple R-group decomposition Composer algorithm superior in many ways for Markush structure generation, but if the Markush core is already known and the main goal is the fragment table generation for the existing molecule library, instead of the creation of Markush, R-group decomposition is the better solution.
Regarding Markush structure generation, the most significant benefits of Markush Composer compared to R-group decomposition the capability to find the scaffold automatically, and capability to create nested R-groups. Due to some conceptual differences, Composer gives better result in most cases. R-group decomposition executes substructure searches with the predefined scaffold against the target molecules one by one, and if the predefined scaffold can fit multiple ways, R-group decomposition simply uses the first hits, this is the base of the decomposition table, and Markush structure can be generated by merging the rows of the decomposition table. In contrast, with it, Markush Composer use all possible hits of multiple automatically recognized and modified MCS results to find the best scaffold and optimal scaffold orientation and it directly generates the Markush structure.
Markush Composer can be easily used from KNIME. The following example illustrating how can you build a simple workflow using the Markush Composer node.
You can download the example project from here Composer.zip.