A Sequence-to-Sequence Transformer Model for Disconnection Aware Retrosynthesis

Retrosynthesis is an approach commonly undertaken when considering the manufacture of novel molecules. During this process, a target molecule is broken down and analyzed by considering the bonds to be changed as well as the functional group interconversion. In modern computer-assisted synthesis planning tools, the predictions of these changes are typically carried out automatically. However there may be some benefit to the decision being guided by those executing the process: typically, chemists have a clear idea where the retrosynthetic change should happen, but not how such a transformation is to be realized. Using a data-driven model, the retrosynthesis task can be further explored by giving chemists the option to explore specific disconnections. In this work, we design an approach to provide this option by adapting a transformer-based model for single-step retrosynthesis. The model takes as input a product SMILES string, in which the atoms where the transformation should occur are tagged accordingly. This model predicts precursors corresponding to a disconnection occurring in the correct location in 88.9% of the test set reactions. The assessment with a forward prediction model shows that 76% of the predictions are chemically correct, with 14.1% perfectly matching the ground truth.

The Medium-to-Short-Term Earthquake Predictions in China and Their Evaluations Based on the R-Score

The earthquake tendency consultations in China, which have been carried out by the China Earthquake Administration for more than 40 yr, are really forward prediction of earthquakes. The results, experiences, and data accumulation are valuable for seismic researches. In this article, the annual, monthly, and weekly predictions produced by the regular earthquake tendency consultations and the rapid postearthquake tendency prediction derived from the irregular ones are presented systematically. In the regular predictions, the areas where earthquakes tend to occur are identified by specific space–time windows. To evaluate the efficiency of the predictions, we apply the R-score method to all the medium-to-short-term efforts. The R-score has been used as a routine tool to test annual predictions in China, in which the hit rate and the percentage of spatial alarms over the whole territory are taken into consideration. Results show that the annual R-scores, during the period of 1990–2020, increased gradually, with the average of 0.293. The examples in 2018 indicate that a considerable proportion of earthquakes with the Ms 5.0 and above were detected by the annual prediction; some earthquakes were detected by the monthly prediction, whereas just only a few earthquakes could be detected by the weekly prediction. The corresponding R-scores are 0.46, 0.11, and 0.002, decreasing obviously with reduction of the prediction time windows, and the smallest one, which is very close to zero, may suggest the minimum time scale for an effective earthquake prediction. We also evaluated efficiency of the irregular predictions by analyzing the practices of 29 Ms≥5.0 earthquakes since January 2019 and found that it is highly possible to do rapid postearthquake tendency prediction in China.

Air-Wake Prediction Based Air-Vehicle Recovery Aids

This report demonstrates the capability of the forward prediction of the properties of the arriving wind at a vessel for time intervals adequate to significantly aid in the recovery of a wide range of air vehicles onto vessels. For craft with flight decks sited in the fore part of the vessel it is adequate to simply predict the arriving wind. For the more difficult task of recovery to stern areas behind superstructure it is also necessary to predict either the explicit properties of the turbulent air-wake or else to predict some quality measure for the aid of recovery under the prevailing conditions. The approach is able to relate the trends in the short-term statistical properties of fluctuating airflow over the flight deck to the trends in the predicted arriving wind.

Predicting Fusarium Head Blight Resistance for Advanced Trials in a Soft Red Winter Wheat Breeding Program With Genomic Selection

Many studies have evaluated the effectiveness of genomic selection (GS) using cross-validation within training populations; however, few have looked at its performance for forward prediction within a breeding program. The objectives for this study were to compare the performance of naïve GS (NGS) models without covariates and multi-trait GS (MTGS) models by predicting two years of F4:7 advanced breeding lines for three Fusarium head blight (FHB) resistance traits, deoxynivalenol (DON) accumulation, Fusarium damaged kernels (FDK), and severity (SEV) in soft red winter wheat and comparing predictions with phenotypic performance over two years of selection based on selection accuracy and response to selection. On average, for DON, the NGS model correctly selected 69.2% of elite genotypes, while the MTGS model correctly selected 70.1% of elite genotypes compared with 33.0% based on phenotypic selection from the advanced generation. During the 2018 breeding cycle, GS models had the greatest response to selection for DON, FDK, and SEV compared with phenotypic selection. The MTGS model performed better than NGS during the 2019 breeding cycle for all three traits, whereas NGS outperformed MTGS during the 2018 breeding cycle for all traits except for SEV. Overall, GS models were comparable, if not better than phenotypic selection for FHB resistance traits. This is particularly helpful when adverse environmental conditions prohibit accurate phenotyping. This study also shows that MTGS models can be effective for forward prediction when there are strong correlations between traits of interest and covariates in both training and validation populations.

Inverse Design Framework with Invertible Neural Networks for Passive Vibration Suppression in Phononic Structures

Automated inverse design methods are critical to the development of metamaterial systems that exhibit special user-demanded properties. While machine learning approaches represent an emerging paradigm in the design of metamaterial structures, the ability to retrieve inverse designs on-demand remains lacking. Such an ability can be useful in accelerating optimization-based inverse design processes. This paper develops an inverse design framework that provides this capability through the novel usage of invertible neural networks (INN). We exploit an INN architecture that can be trained to perform forward prediction over a set of high-fidelity samples, and automatically learns the reverse mapping with guaranteed invertibility. We apply this INN for modeling the frequency response of periodic and aperiodic phononic structures, with the performance demonstrated on vibration suppression of drill pipes. Training and testing samples are generated by employing a Transfer Matrix Method. The INN models provide competitive forward and inverse prediction performance compared to typical deep neural networks (DNN). These INN models are used to retrieve approximate inverse designs for a queried non-resonant frequency range; these inverse designs are then used to initialize a constrained gradient-based optimization process to find a more accurate inverse design that also minimizes mass. The INN initialized optimizations are found to be generally superior in terms of the queried property and mass compared to randomly-initialized and inverse DNN-initialized optimizations. Particle Swarm Optimization with INN-derived initial points is then found to provide even better solutions, especially for the higher-dimensional aperiodic structures.

Molecular Transformer-aided Biocatalysed Synthesis Planning

Enzyme catalysts are an integral part of green chemistry strategies towards a more sustainable and resource-efficient chemical synthesis. However, the use of enzymes on unreported substrates and their specific stereo- and regioselectivity are domain-specific knowledge factors that require decades of field experience to master. This makes the retrosynthesis of given targets with biocatalysed reactions a significant challenge. Here, we use the molecular transformer architecture to capture the latent knowledge about enzymatic activity from a large data set of publicly available biochemical reactions, extending forward reaction and retrosynthetic pathway prediction to the domain of biocatalysis. We introduce the use of a class token based on the EC classification scheme that allows to capture catalysis patterns among different enzymes belonging to the same hierarchical families. The forward prediction model achieves an accuracy of 49.6% and 62.7%, top-1 and top-5 respectively, while the single-step retrosynthetic model shows a round-trip accuracy of 39.6% and 42.6%, top-1 and top-10 respectively. Trained models and curated data are made publicly available with the hope of promoting enzymatic catalysis and making green chemistry more accessible through the use of digital technologies.

Molecular Transformer-aided Biocatalysed Synthesis Planning

Enzyme catalysts are an integral part of green chemistry strategies towards a more sustainable and resource-efficient chemical synthesis. However, the use of enzymes on unreported substrates and their specific stereo- and regioselectivity are domain-specific knowledge factors that require decades of field experience to master. This makes the retrosynthesis of given targets with biocatalysed reactions a significant challenge. Here, we use the molecular transformer architecture to capture the latent knowledge about enzymatic activity from a large data set of publicly available biochemical reactions, extending forward reaction and retrosynthetic pathway prediction to the domain of biocatalysis. We introduce the use of a class token based on the EC classification scheme that allows to capture catalysis patterns among different enzymes belonging to the same hierarchical families. The forward prediction model achieves an accuracy of 49.6% and 62.7%, top-1 and top-5 respectively, while the single-step retrosynthetic model shows a round-trip accuracy of 39.6% and 42.6%, top-1 and top-10 respectively. Trained models and curated data are made publicly available with the hope of promoting enzymatic catalysis and making green chemistry more accessible through the use of digital technologies.