A Reaction Based Combinatorial Library Generation System

- Implemented with Java and the Daylight Reaction Toolkit

TOC

Library Synthesis Examples

Software Requirements

Commercial Combinatorial Chemistry Software

Library Generation Analysis

Daylight Reaction Toolkit Implementation

Demo

In-depth Discussion

Java Implementation

Asking Daylight

Acknowledgment

Library Synthesis Examples

• Add one precursor list per step (Figure 1)

• Multiple precursor lists added together in one reaction step (Figure 2)

• The product library of special interpretation (Figure 3)
• A closer look at step 2 (Figure 4)
• Product libraries (Figure 5)
• Reusable reactions (Figure 6)

 
Top

Software Requirements

Chemistry

• Perform correct chemical transformation
• Able to identify main product
• Perform a set of pre-defined generic reactions, for example, acid de-protection

Visualization

• Structure and transformation depiction
• Generic transformation definition
• Multiple synthesis steps depcition

System Integration

• Network computing
• Cross-platform
• Other information system integration

Top

Commercial Combinatorial Chemistry Software

MDL

• Product oriented
• Can not handle some reaction precursors to mixture synthesis library generation (Figure 3)
• Does not handle UGI reaction library (Figure 2) elegantly
• Does not perform a set of pre-defined generic reactions (Figure 6)

Tripos

• Reaction based, but buggy

ChemX

• Reaction/Product oriented, but generic reactions can not be reused, and buggy
• Does not perform a set of pre-defined generic reactions (Figure 6)

Daylight

• Product oriented, very difficult for a bench chemist to use
• Does not perform a set of pre-defined generic reactions (Figure 6)

All of the above software are isolated from other information systems, and they are not cross-platform applications.

Top

Library Generation Analysis

• Parallel Synthesis OOA (Figure 7)
• Mixture Synthesis OOA (Figure 8)

Top

Daylight Reaction Toolkit Implementation

SMIRKS is a language that describes a generic TRANSFORMATION. It does not have to reflect a chemical reaction mechanism. The output (products) of a transformation is completely user-designed.
 

• Parallel synthesis program can
• loop through reactant arrays, perform transformation on them to yield products,
• identify main product,
• preserve the stereochemistry from reactants to products,
• handle multiple SMIRKS in one transformation step,
• track multiple step synthesis,
• handle multiple library synthesis.
• Mixture synthesis program can
• clip precursors and make new core,
• identify main product,
• preserve the stereochemistry from reactants to products if the clipping does not happen to the stereocenter,
• handle multiple SMIRKS in one transformation step,
• track multiple step synthesis and provides total synthesis scheme,
• handle multiple library synthesis.

Top

Demo

Parallel Synthesis

• Add one precursor list per step
• step 1 (Figure 9)
• step 2 (Figure 10)
• step 3 (Figure 11)
• Multiple precursor lists added together in one reaction step (Figure 12)
• The product libary of special interpretation
• step 1 (Figure 13)
• step 2 (Figure 14)
• step 3 (Figure 15)
• step 4 (Figure 16)
• Reusable Reactions (Figure 17)

Mixture Synthesis

• Add one precursor list per step
• step 1, before transformation (Figure 18)
• step 1, after transformation (Figure 19)
• step 2, before transformation (Figure 20)
• step 2, after transformation (Figure 21)
• step 3, after transformation (Figure 22)
• Multiple precursor lists added together in one reaction step
• before transformation (Figure 23)
• after transformation (Figure 24)
• The product libary of special interpretation
• step 1, before transformation (Figure 25)
• step 1, after transformation (Figure 26)
• step 2, after transformation (Figure 27)
• step 3, before transformation (Figure 28)
• step 3, after transformation (Figure 29)
• step 4, after transformation (Figure 30)

Top

In-depth Discussion

Parallel Synthesis Transformation

• no need to specify every atom in reactants or products
• need to specify bond breaking and formation
• need to specify atomic environment around the bond that undergoes change
• need to have atom-atom mapping for those atoms that defined the bond that undergoes change

Examples

• Generic transformation for Figure 1, step 1 -- add one precursor list in a reaction step (Figure 31)

• Generic transformation for Figure 2 -- add 4 precursor lists together (Figure 32)

• Generic transformation for Figure 3, Step 3 -- make variable size rings (Figure 33)

Mixture Synthesis Transformation

The common structure feature, which the elements in a precursor list share, will be abstracted to the library core after a transformation. The structural diverse part that the elements in a precursor list do not share has to be represented to have the common structural feature clipped off. Therefore, in addition to the parallel synthesis SMIRKS specification, attachment points of a precursor list to the core have to be specified in mixture synthesis generic transformation. "*" is used in generic transformation as attachment points. Elemental symbols [Tl], [Po], [Pt], [Bi], etc. are used in reactant and product SMILES to indicate the attachments. When the structures containing attachment points are shown to user, the elemental symbols that indicate the attachment points are replaced by A1, A2, A3, etc..

Examples

• Generic transformation for Figure 1, step 1 -- clipped precursor has one attachment point to the core
• User defined generic transformation (Figure 34)
• Program derived precursor 1 clipping transformation (Figure 35)
• Program derived precursor 2 clipping transformation (Figure 36)
• Make core (Figure 37)
• Generic transformation for Figure 3, step 3 -- clipped precursor has two attachment points to the core
• User defined generic transformation (Figure 38)
• Program derived precursor 1 clipping transformation (Figure 39)
• Program derived uniting transformation for clipped off parts from precursor 1 (Figure 40)
• Program derived precursor 2 clipping transformation (Figure 41)
• Program derived uniting transformation for clipped off parts from precursor 2 (Figure 42)
• Make core (Figure 43)

Main Product Identification

• Make sure the generic structure of the main product has features different from those of balance pieces,
• In the current application, the main product SMARTS is right after the '>>' symbol in a transformation SMIRKS,
• Loop through transformation product molecules, only those who match (SSS) the main product generic structure are the main products.
• If exactly what are by-products is of concern, make sure the balance pieces in SMIRKS reflect the correct by-products.

Top

Java Implementation

Application Development Framework

• Model/View/Controller (MVC)
• Observer/Observable

Chemical Structure Depiction

• depict a chemical structure from a molfile,
• color atoms,
• show attachment atoms [At], [Po], [Pb], etc. as A1, A2, A3, etc.,
• zoom structure.

Network Communication Via RMI

• Java client can run on any platform that supports Java 1.1.
• Server (Java + c) resides on the machine that has Daylight license.
• Java RMI handles the client-server communication.

Top

Asking Daylight

• to provide SMIRKS and structure query tool in Java -- a Java version of GRIN, which can be subclassed.
• to generate better structure coordinates for SMILES.
• to provide SMIRKS with full SMARTS features (allowing recursive SMARTS).
• to allow non-elemental symbols in SMILES, or allow subclass SMILES.

Acknowledgment

Michael Jenkins (jenkinsms@abbott.com)
Wade Warren, Ph.D. (wadew@abbott.com)
Joe Cesarone, Ph.D. (cesaronej@abbott.com)

Top