Leveraging Language Model Multitasking To Predict C-H Borylation Selectivity.

C-H borylation is a high-value transformation in the synthesis of lead candidates for the pharmaceutical industry because a wide array of downstream coupling reactions is available. However, predicting its regioselectivity, especially in drug-like molecules that may contain multiple heterocycles, is not a trivial task. Using a data set of borylation reactions from Reaxys, we explored how a language model originally trained on USPTO_500_MT, a broad-scope set of patent data, can be used to predict the C-H borylation reaction product in different modes: product generation and site reactivity classification. Our fine-tuned T5Chem multitask language model can generate the correct product in 79% of cases. It can also classify the reactive aromatic C-H bonds with 95% accuracy and 88% positive predictive value, exceeding purpose-developed graph-based neural networks.

Comparative Genomic Analysis of the Pattern of Evolution of Male and Female Reproductive Proteins in Seed Beetles.

Male seminal fluid proteins often show signs of positive selection and divergent evolution, believed to reflect male-female coevolution. Yet, our understanding of the predicted concerted evolution of seminal fluid proteins and female reproductive proteins is limited. We sequenced, assembled, and annotated the genome of two species of seed beetles allowing a comparative analysis of four closely related species of these herbivorous insects. We compare the general pattern of evolution in genes encoding seminal fluid proteins and female reproductive proteins with those in digestive protein genes and well-conserved reference genes. We found that female reproductive proteins showed an overall ratio of nonsynonymous to synonymous substitutions (ω) similar to that of conserved genes, while seminal fluid proteins and digestive proteins exhibited higher overall ω values. Further, seminal fluid proteins and digestive proteins showed a higher proportion of sites putatively under positive selection, and explicit tests showed no difference in relaxed selection between protein types. Evolutionary rate covariation analyses showed that evolutionary rates among seminal fluid proteins were on average more closely correlated with those in female reproductive proteins than with either digestive or conserved genes. Gene expression showed the expected negative covariation with ω values, except for male-biased genes where this negative relationship was reversed. In conclusion, seminal fluid proteins showed relatively rapid evolution and signs of positive selection. In contrast, female reproductive proteins evolved at a lower rate under selective constraints, on par with genes known to be well conserved. Although our findings provide support for concerted evolution of seminal fluid proteins and female reproductive proteins, they also suggest that these two classes of proteins evolve under partly distinct selective regimes.