Search for: All records

Abstract Herein, we describe our synthetic efforts toward the pupukeanane natural products, in which we have completed the first enantiospecific route to 2‐isocyanoallopupukeanane in 10 steps (formal synthesis), enabled by a key Pd‐mediated cyclization cascade. This subsequently facilitated an unprecedented bio‐inspired “contra‐biosynthetic” rearrangement, providing divergent access to 9‐isocyanopupukeanane in 15 steps (formal synthesis). Computational studies provide insight into the nature of this rearrangement. 

Abstract A synthetic method for the palladium‐catalyzed cyanation of aryl boronic acids using bench stable and non‐toxicN‐cyanosuccinimide has been developed. High‐throughput experimentation facilitated the screen of 90 different ligands and the resultant statistical data analysis identified that ligand σ‐donation, π‐acidity and sterics are key drivers that govern yield. Categorization into three ligand groups – monophosphines, bisphosphines and miscellaneous – was performed before the analysis. For the monophosphines, the yield of the reaction increases for strong σ‐donating, weak π‐accepting ligands, with flexible pendant substituents. For the bisphosphines, the yield predominantly correlates with ligand lability. The applicability of the designed reaction to a wider substrate scope was investigated, showing good functional group tolerance in particular with boronic acids bearing electron‐withdrawing substituents. This work outlines the development of a novel reaction, coupled with a fast and efficient workflow to gain understanding of the optimal ligand properties for the design of improved palladium cross‐coupling catalysts. 

The lack of publicly available, large, and unbiased datasets is a key bottleneck for the application of machine learning (ML) methods in synthetic chemistry. Data from electronic laboratory notebooks (ELNs) could provide less biased, large datasets, but no such datasets have been made publicly available. The first real-world dataset from the ELNs of a large pharmaceutical company is disclosed and its relationship to high-throughput experimentation (HTE) datasets is described. For chemical yield predictions, a key task in chemical synthesis, an attributed graph neural network (AGNN) performs as well as or better than the best previous models on two HTE datasets for the Suzuki–Miyaura and Buchwald–Hartwig reactions. However, training the AGNN on an ELN dataset does not lead to a predictive model. The implications of using ELN data for training ML-based models are discussed in the context of yield predictions.