CICC — Chemical Informatics And Cyberinfrastructure Collaboratory Department

CICC - Chemical Informatics And Cyberinfrastructure Collaboratory Department of Chemistry & School of Informatics Indiana University Bloomington Varuna: An Integrated Modeling Environment and Database for Quantum Chemical Simulations Chemical Prototype Projects October 21, 2005 Mu-Hyun Baik

State of Affairs in Computational Chemistry n High-level quantum simulations based on Density Functional Theory allow for very reliable simulations of chemical reactions for systems containing up to 500 atoms. n Combining Quantum Mechanics and Molecular Mechanics, we can construct highly realistic computer models of biologically relevant reactions. n Currently, chemical modeling studies are done in an isolated fashion and the computed data is typically collected in an unorganized manner (directory-jungle) and disregarded after completion of the study. n Modeling is currently done manually: vi, emacs and ssh are currently the most common interfaces of computational chemists. 2

Cyberinfrastructure Development n Depository for computational chemistry data. q q n Automated data collection and categorization Chemical structure recognition Mining of quantum chemical data User independent domain expertise Development of an integrated modeling environment q Services: Automated execution of calculations n q q Automatic generation of input files, communication with number crunchers, recognition and correction of typical failures, automated import of main results, etc. Computational resource management Visualization 3

Data Structure Currently Implemented: - Metadata: QM parameters, Project data - Results: Energy components - Parser extracts all important results - Visualizations Future Work: - Structure recognition (2 D and 3 D fingerprints, SMILES, etc…. ) - Automatic generation of new structures based on computed results 4

Automated Computational Chemistry - Increase efficiency through automation => Make life easier - Allow high-throughput production => Combinatorial Computational Chemistry - Increase depth of wavefunction analysis => Automated pattern-search - Simplify and visualize complicated data in intuitive graphical representations - Allow information recycling => Accumulation of group expertise (Data depository system, Web-Interface) 5

Chemical Prototype Projects 6

Pathogenesis of Alzheimer’s Disease Neuritic plaque with a core made of Cu-b-Amyloid complex AD with cortical atrophy 7

8

How does Varuna fit into all this? n Force-field Database: q q Currently, Cu-Ligand force-fields are being generated manually. We would like to develop a Service component that will do this automatically These force-fields will be made available in the database. n We already have ~400 plausible Cu-b-Amyloid high-resolution structures with QM energies: Data Mining Services are needed to compare structural similarities, reactivity indices, etc. n The reactivity of the Cu-center in the peptide must be compared systematically against small molecule models. 9

Immediate Challenges n A 3 D structural representation is needed that can deal with: q q n Non-integer bond-orders, transition state structures with multi-center/multi-electron bonds Many different quantum chemically derived property topologies The metadata is complex because of many technical parameters that make calculations difficult to compare 10

Cisplatin: Profiling an Anticancer Drug 11

Computational Organic Chemistry 12

Diastereoselective [4+2+2] Carbocyclization - What is the mechanism of this transformation? - What is the source of the diastereoselectivity? - Can the scope of the reaction be extended? - Can we reverse the stereo-control using the same methodology? Evans, P. A. et al. Chem. Commun. 2005, 63 13

Who cares ? Mehta, Singh. Chem. Rev. 1999, 881 14

Reaction Energy Profiles Low CO Pressure Low diastereoselectivity High CO Pressure High diastereoselectivity 15

Collaborative Network CICC Center for Catalysis (IU) Caulton Mindiola Evans Johnston Williams Baik-Group (IU) Computational Chemistry Molecular Modelling Lippard (MIT) Cisplatin, Methane Monooxygenase Jacobsen (Harvard) Asymmetric Catalysis, Enzymatic Oxidations Newcomb (UI-Chicago) B 12 -Dependent Enzymes Szalai (UMBC) Alzheimer’s Disease Sames (Columbia) Ir-, Rh-Catalyzed C-H activation 16

Center for Catalysis at IU-Bloomington Organic Synthesis Andy Evans Jeff Johnston Molecular Modeling Mookie Baik Organometallic Catalyst Design Dan Mindiola Ken Caulton Rational Design of Well-Defined, Efficient and mechanistically fully understood Catalysts for Natural Product Synthesis, Polymerization and C-C/C-H activation. Educational Goal: A new breed of chemists who can conduct high -level research in all three areas of Organic, Inorganic and Computational/Theoretical Research 17

18

General Research Philosophy Experiments New Chemistry Structures, Lifetimes, Rates, Isotope-Effects Activation Enthalpies, Redox-Potentials…. Prediction Model Chemistry HOW? Theoretical Tools DFT, MP 2, MM, QM/MM, etc. . Model Chemistry WHY? Analysis Chemical Intuition MO-Diagram Energy-Decomposition What-If Game Handwaving 19

Inherent Problems of Organic Mechanism Discovery n Most of the time all you have is a reactant and a product, if you are lucky. n Intermediates, particularly the interesting reactive ones, can’t be observed directly. n “Classical Approach” of Constructing a New Mechanism: q Memorize as many as possible known mechanisms q Try to recognize similarities (mostly structural) and assume that worked for one reaction may work for another n Mechanisms are often quite “arbitrary”. 20

“Classical” Approach to Proposing a Mechanism What we’ve seen before: Pauson-Khand-type Reaction Evans, P. A. et al. J. Am. Chem. Soc. 2001, 123, 4609 Magnus, P. et al. Tetrahedron 1985, 41, 5861 Buchwald S. L. et al. J. Am. Chem. Soc. 1996, 118, 11688. 21

“Classical” Approach to Proposing a Mechanism “Logical” mechanism for the [4+2+2]: Stereocontrol: Rh coordination is facially selective. The sterically bulky R 1 group directs Rh to the correct side of the p-component. Evans, P. A. et al. Chem. Commun. 2005, 63 22

Let’s think about this…. - Oxidative Addition involving the triple bond should be facile. => (A) and (B) can’t be rate determining! - So, forming either bond (A) or (B) first is plausible, but: - Form (B) first => Stereochemistry at C 2 is fixed !! - Stereocontrol at a reaction Step that is NOT rate determining? ? 23

New Proposal J. Am. Chem. Soc. 2005, 127, 1603 24

Computational Model Chemistry - Density Functional Theory @ B 3 LYP/cc-p. VTZ(-f) (Jaguar) - Numerically efficient up to 300 atoms => no compromises with respect to Model Size 25

Entropy 26

Continuum Solvation Model 27

Computed Reaction Energy Profiles J. Am. Chem. Soc. 2005, 127, 1603 28

Computed Reaction Energy Profiles J. Am. Chem. Soc. 2005, 127, 1603 29

Diastereoselectivity ? ? J. Am. Chem. Soc. 2005, 127, 1603 30

Reason for Diastereoselectivity J. Am. Chem. Soc. 2005, 127, 1603 31

Understanding Pauson-Khand-Type Reactions: [2+2+1] 32

Mechanistic Alternatives Low CO pressure High CO pressure 33

What about Structural Alternatives? 34

Reaction Energy Profiles Low CO Pressure Low diastereoselectivity High CO Pressure High diastereoselectivity 35

Why is this reaction diastereoselective? Partial Charge Analysis Syn-Product forms by (+)-directed polarization. Anti-Product forms by (-)-directed polarization. 36

What is the physical basis of the new rule? 37

What is the physical basis of the new rule? 38

But, can we predict new chemistry? n Diastereoselectivity is CO-pressure dependent! 39

Precision in the Eyes of an Organic Chemist dppp: 1, 3 -bis(diphenylphosphino)propane 40

Hey – who said anything about phosphine? 41

So, WHY is this happening? Low CO Pressure Low diastereoselectivity High CO Pressure High diastereoselectivity 42

Does this make sense NOW? dppp: 1, 3 -bis(diphenylphosphino)propane 43

More Predictions Will Electron withdrawing groups on R 1 reverse ds ? ? Target: No! But: Can’t be made? 44

Conclusions n Theoretical “Characters” can actually predict new stuff if they try hard. n The diastereoselectivity of Rh-catalyzed Pauson-Khand reaction is a rare example of a purely electronically driven stereo-control (close to no steric influence!). n “Spectator Ligands” are actually not really just spectators at all. n Organic Chemistry does not necessarily have to be synonymous with: Alchemy or Mindless Memorizing 45

Center for Catalysis at IU-Bloomington Organic Synthesis Andy Evans Jeff Johnston Molecular Modeling Mookie Baik Organometallic Catalyst Design Dan Mindiola Ken Caulton Rational Design of Well-Defined, Efficient and mechanistically fully understood Catalysts for Natural Product Synthesis, Polymerization and C-C/C-H activation. Educational Goal: A new breed of chemists who can conduct high -level research in all three areas of Organic, Inorganic and Computational/Theoretical Research 46

47