Asymmetric Synthesis DOI: 10.1002/ange.200704774 Toward a Computational Tool Predicting the Stereochemical Outcome of Asymmetric Reactions: Development and Application of a Rapid and Accurate Program Based on Organic Principles** ChristopherR.Corbeil,SabineThielges,JeremyA.Schwartzentruber,andNicolasMoitessier* Asymmetric catalyst discovery as currently practiced often relies on expensive, and sometimes serendipitous, stepwise optimization and/or library screening. [1] We believe that this paradigm is poised to change, as computational predictive methodshavereachedalevelofaccuracythatobviatesmany steps now done manually. We report herein the early version of a new program, ACE (asymmetric catalyst evaluation), its underlying concepts, and the assessment of its applicability and accuracy in distinguishing efficient asymmetric catalysts or chiral auxiliaries from inferior ones. Although much effort has been directed toward the development of computer-aided drug design tools, there has been little investigation into computational tools for asym- metric catalyst design. Nowadays, the fields of quantum mechanics(QM)andquantummechanics/molecularmechan- ics (QM/MM) [2] are highly developed and have yielded accurate predictions of asymmetric reaction stereoselectivi- ties. [3–6] However, QM methods would require months of computation to screen a library of potential catalysts in the searchfornewones.Toaddressthisissue,othermethodswere developed,whichincludereversedocking [7,8] andquantitative structure–selectivity relationships [9–11] and more specifically the use of quantum mechanics interaction fields. [12,13] As another alternative to QM techniques, molecular mechanics applied to ground-state structures have been used. [14] Advanced MM-based transition-state (TS) techniques, which accurately predict TS structures and their relative potentialenergies,havealsobeenreported. [15] Althoughthese methods (e.g., Q2MM (QM-guided MM), [16] using TSFF (transition-state force fields), [17] SEAM (seam of two poten- tial-energy functions), [18,19] EVB (empirical valence bond), [20,21] and MCMM (multiconfiguration MM) [22] ) have showngreatpotentialinlocatingandinvestigatingTSs,onlya very few studies were reported that attempted to predict the stereochemical outcome of reactions, [7,8,14,23–28] with even fewer applications to the design of new asymmetric cata- lysts. [13,29,30] In fact, one major shortcoming of force fields is the lack of accurate parameters for metal complexes, which arenecessarytomodelmetal-catalyzedreactionsandneedto be specifically developed. [31] ACE is a molecular-mechanics-based independent pro- gramthathasbeendevelopedfromsimpleorganicchemistry principles. For example, the Hammond–Leffler postulate states that the TS is most similar to the species (reactants or products) which it is closest to in energy. Following this principle, ACE constructs TSs from a linear combination of reactants and products, including a factor l describing the position of the TS on the potential-energy surface [Eq. (1), TS ¼ð1lÞ reactant þ l product ð1Þ 0 < l < 1]. A similar approach is used to locate transition states by the EVB method mentioned above, in which l is changed stepwise from 0 to 1 to find the maximum energy correspondingtotheTS.EVBhasindeedbeenusedsuccess- fully in the study of several enzymatic mechanisms. [21] Within ACE, interactions between two atoms forming a bond are described as both covalent-bond and nonbonded interactions with weights (1l) and l for each of these two types of interactions. Angles, torsion angles, and nonbonded interac- tions between atoms of the reacting center are also scaled by either (1l) if found in the reactants or l if found in the products. As a comparison, l can be related to the Brønsted coefficient, which measures the role of the reacting partners inaTS. As stated by Curtin and Hammett, stereomeric excesses can be derived from the difference in the diastereomeric TS energies,inthiscasetheMM3*forcefieldpotentialenergies. This force field has already been used with the SEAM and TSFFapproaches to predict TS energy differences. Toassesstheaccuracyofourmethod,weinvestigatedthe asymmetricDiels–Aldercycloadditionusingchiralauxiliaries (Scheme1). For this purpose, 44 systems were selected from the literature involving varying dienes and dienophiles. For each of the diene–dienophile pairs, reactants and productswerebuiltconsideringonlyan endo attack,knownto be favored in this type of reaction. Prior to running the computation, l mustbeset.ItiswellknownthatDiels–Alder reactions in the presence of strong Lewis acids have low energies of activation and early TSs, a situation which corresponds to low values of l. In order to evaluate the impactoftheselected l value,valueswereusedrangingfrom 0.10 to 0.60 in steps of 0.10. [*] C. R. Corbeil, S. Thielges, J. A. Schwartzentruber, Prof. Dr. N. Moitessier Department of Chemistry, McGill University 801 Sherbrooke Street W., MontrØal, QuØbec H3A 2K6 (Canada) Fax: (+ 1)514-398-3797 E-mail: nicolas.moitessier@mcgill.ca Homepage: moitessier-group.mcgill.ca [**] We thank the Canadian Foundation for Innovation for financial support through the New Opportunities Fund program, NSERC, ViroChem Pharma, FQRNT, and CIHR for financial support and RQCHP for generous allocation of computer resources. Supporting information for this article is available on the WWW under http://www.angewandte.org or from the author. Angewandte Chemie 2675 Angew. Chem. 2008, 120, 2675 –2678 # 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim