RESUMEN
This work presents efforts to augment the performance of data-driven machine learning algorithms for reaction template recommendation used in computer-aided synthesis planning software. Often, machine learning models designed to perform the task of prioritizing reaction templates or molecular transformations are focused on reporting high-accuracy metrics for the one-to-one mapping of product molecules in reaction databases to the template extracted from the recorded reaction. The available templates that get selected for inclusion in these machine learning models have been previously limited to those that appear frequently in the reaction databases and exclude potentially useful transformations. By augmenting open-access data sets of organic reactions with explicitly calculated template applicability and pretraining a template-relevance neural network on this augmented applicability data set, we report an increase in the template applicability recall and an increase in the diversity of predicted precursors. The augmentation and pretraining effectively teaches the neural network an increased set of templates that could theoretically lead to successful reactions for a given target. Even on a small data set of well-curated reactions, the data augmentation and pretraining methods resulted in an increase in top-1 accuracy, especially for rare templates, indicating that these strategies can be very useful for small data sets.
Asunto(s)
Redes Neurales de la Computación , Programas Informáticos , Algoritmos , Computadores , Aprendizaje AutomáticoRESUMEN
Explicit water atomistic molecular dynamics simulations (200 ns, â¼330,000 atoms) were performed to study the effects of galactosylation in the Fc domain of immunoglobulin G1. Two glycoforms were simulated to observe changes in protein-carbohydrate interactions and carbohydrate structure. A high degree of flexibility was observed in the small hinge region of the protein, while large domains remained stable. The hinge region flexibility allowed both translation and rotation of the domains relative to each other, resulting in a large number of possible conformations available. The distributions of rotational orientations between the Fab1 and Fab2 domains showed that while these domains are able to orient themselves rather freely pointing in space they rotated in unison to remain rotationally oriented at specific angles. Additionally, removing specific terminal galactose residues increased the mobility of the carbohydrate, resulting in different protein-carbohydrate interactions. Glycosylation has been suggested as a route to improve the aggregation resistance of monoclonal antibodies for therapeutic treatments to aid the immune system. The results presented here may provide insight into the search for IgG molecules with increased aggregation resistance to be used as monoclonal antibodies.