Title

LHASA Group, Department of Chemistry and Chemical Biology, Harvard University

Alan Long, Alan Long, RongRong Chen, Craig Chen, Craig MarbyMarby, , Alex Alex SukharevskySukharevsky, Keith Ohm, Keith Ohm

Synthesis PlanningCompound Library Generation

From

To

Applications of LHASA TechnologyApplications of LHASA Technology

OutlineOutline

Target-oriented SynthesisClassical LHASANon-interactive LHASA

Diversity-oriented SynthesisLCOLI = LHASA for Compound Libraries

Classical LHASA ProcessingClassical LHASA Processing

Using LHASA EffectivelyUsing LHASA Effectively

• LHASA knowledge base organization- Goal Transforms, Subgoal Transforms- Tactical Combinations, Long-range Transforms

• LHASA transform keying mechanisms- Functional groups- One-dimensional patterns- Two-dimensional patterns

• LHASA transform selection strategies and tactics:- Short-range: unconstrained, disconnective, reconnective, etc.- Topological: cyclic strategic, acyclic strategic, etc. - Long-range: Diels-Alder, Robinson annulation, olefins, etc.- Stereochemical: diastereoselective, enantioselective, etc.- Starting-material-based: aromatic pool, chiral pool, etc.

Too many Decisions!Too many Decisions!

Automatic route growth in LHASAAutomatic route growth in LHASA

• Automatic “best-sequence” generation- Maximize retrosynthetic simplification- Maximize overall sequence rating

• “Find-Best-Solution” Button:

Process Ordering:

Complexity

Sequence Rating

NonNon--interactive LHASA interactive LHASA 1 of 21 of 2

Process Constraints:

Current candidates for

processing

Target

Complexity

Sequence Rating

Simple enough(obtain commercially orknown compounds)

Prevents non-productive (non-simplifying) routes from being generated

Min Max

Max = 1.0

Discards unpromising sequences

Min

Processing Threshold Function

NonNon--interactive LHASA interactive LHASA 2 of 22 of 2

LCOLI LCOLI –– LHASA for Compound LibrariesLHASA for Compound Libraries

• Uses existing LHASA knowledge base

• Uses 2-D patterns to create mechanism commands leading to a virtual product

O H O O H O“ | “ “ | “

C,H,O,N-C-C-C-C-C-C,H,O,N => C,H,O,N-C-C=C + C-C-C,H,O,N

• Runs virtual product through LHASA and compares virtual precursor with original reactants

LCOLI Processing LCOLI Processing –– Library GenerationLibrary Generation

Transform 153Michael Reaction

• text here

Simple Combinatorial LibrarySimple Combinatorial Library

• Can use multi-component reactions to maximize diversity

Example: Ugi four-component alpha-aminoacyl amide synthesis

LCOLI LCOLI –– Maximizing DiversityMaximizing Diversity

• More LCOLI output

LCOLI Processing LCOLI Processing –– Four starting material filesFour starting material files

LCOLI Processing AnalysisLCOLI Processing Analysis

Killed combinations:

1. Pattern mismatches:

Amino Acids such as PROLINE contain secondary amino groups instead of the primary amines required by the reaction pattern …

2. Interfering functionality:

3. Adverse structural features:

Number12

x 12x 12x 20

34,560

Building Blocks

AlcoholsAldehydes

IsocyanidesAmino Acids

Combinations =

Result

30,702Ugi Products

Transform 1934

Ugi Reaction with Amino Acids

LCOLI Processing LCOLI Processing –– Toxicology screeningToxicology screening

• LHASA Retrosynthetic Analysis

• APSO Teaching of organic synthesis

• PROTECT Functional group protection

• DEREK Toxicology prediction

• LCOLI Compound library generation

The LHASA SuiteThe LHASA Suite

Conclusions

• Generate promising retrosynthetic routesautomatically, with control over depth vs. breadth

• Create and screen virtual combinatorial libraries• Provide framework for future development

Next Steps

• Integrate and centralize LHASA applications • Develop Visual CHMTRN Tool• Extend application of LHASA knowledge base technology

Conclusions and Next StepsConclusions and Next Steps

AcknowledgementsAcknowledgements