High-Level Data Fusion Enables the Chemoinformatically Guided Discovery of Chiral Disulfonimide Catalysts for Atropselective Iodination of 2-Amino-6-arylpyridines.

The atropselective iodination of 2-amino-6-arylpyridines catalyzed by chiral disulfonimides (DSIs) is described. Key to the development of this transformation was the use of a chemoinformatically guided workflow for the curation of a structurally diverse training set of DSI catalysts. Utilization of this catalyst training set in the atropselective iodination across a variety 2-aminopyridine substrates allowed for the recommendation of statistically higher-performing DSIs for this reaction. Data Fusion techniques were implemented to successfully predict the performance of catalysts when classical linear regression analysis failed to provide suitable models. This effort identified a privileged class of 3,3'-alkynyl-DSI catalysts which were effective in catalyzing the iodination of a variety of 2-amino-6-arylpyridines with high stereoselectivity and generality. Subsequent preparative-scale demonstrations highlighted the utility of this reaction by providing iodinated pyridines >90:10 er and in good chemical yield.

A Unified Strategy for the Asymmetric Synthesis of Highly Substituted 1,2-Amino Alcohols Leading to Highly Substituted Bisoxazoline Ligands.

A general procedure for the asymmetric synthesis of highly substituted 1,2-amino alcohols in high yield and diastereoselectivity is described that uses organometallic additions of a wide range of nucleophiles to tert-butylsulfinimines as the key step. The addition of organolithium reagents to these imines follows a modified Davis model. The diastereoselectivity for this reaction depends significantly on both the nucleophile and electrophile. These highly substituted 1,2-amino alcohols are used to synthesize stereochemically diverse and structurally novel, polysubstituted 2,2'-methylene(bisoxazoline) ligands in high yields.

Development of a Computer-Guided Workflow for Catalyst Optimization. Descriptor Validation, Subset Selection, and Training Set Analysis.

Modern, enantioselective catalyst development is driven largely by empiricism. Although this approach has fostered the introduction of most of the existing synthetic methods, it is inherently limited by the skill, creativity, and chemical intuition of the practitioner. Herein, we present a complementary approach to catalyst optimization in which statistical methods are used at each stage to streamline development. To construct the optimization informatics workflow, a number of critical components had to be subjected to rigorous validation. First, the critically important molecular descriptors were validated in two case studies to establish the importance of conformation-dependent molecular representations. Next, with a large data set available, it was possible to investigate the amount of data necessary to make predictive models with different modeling methods. Given the commercial availability of many catalyst structures, it was possible to compare models generated with algorithmically selected training sets and commercially available training sets. Finally, the augmentation of limited data sets is demonstrated in a method informed by unsupervised learning to restore the accuracy of the generated models.

Prediction of higher-selectivity catalysts by computer-driven workflow and machine learning.

Catalyst design in asymmetric reaction development has traditionally been driven by empiricism, wherein experimentalists attempt to qualitatively recognize structural patterns to improve selectivity. Machine learning algorithms and chemoinformatics can potentially accelerate this process by recognizing otherwise inscrutable patterns in large datasets. Herein we report a computationally guided workflow for chiral catalyst selection using chemoinformatics at every stage of development. Robust molecular descriptors that are agnostic to the catalyst scaffold allow for selection of a universal training set on the basis of steric and electronic properties. This set can be used to train machine learning methods to make highly accurate predictive models over a broad range of selectivity space. Using support vector machines and deep feed-forward neural networks, we demonstrate accurate predictive modeling in the chiral phosphoric acid-catalyzed thiol addition to N-acylimines.