Prediction of chemical reaction yields with large-scale multi-view pre-training.

Developing machine learning models with high generalization capability for predicting chemical reaction yields is of significant interest and importance. The efficacy of such models depends heavily on the representation of chemical reactions, which has commonly been learned from SMILES or graphs of molecules using deep neural networks. However, the progression of chemical reactions is inherently determined by the molecular 3D geometric properties, which have been recently highlighted as crucial features in accurately predicting molecular properties and chemical reactions. Additionally, large-scale pre-training has been shown to be essential in enhancing the generalization capability of complex deep learning models. Based on these considerations, we propose the Reaction Multi-View Pre-training (ReaMVP) framework, which leverages self-supervised learning techniques and a two-stage pre-training strategy to predict chemical reaction yields. By incorporating multi-view learning with 3D geometric information, ReaMVP achieves state-of-the-art performance on two benchmark datasets. Notably, the experimental results indicate that ReaMVP has a significant advantage in predicting out-of-sample data, suggesting an enhanced generalization ability to predict new reactions. Scientific Contribution: This study presents the ReaMVP framework, which improves the generalization capability of machine learning models for predicting chemical reaction yields. By integrating sequential and geometric views and leveraging self-supervised learning techniques with a two-stage pre-training strategy, ReaMVP achieves state-of-the-art performance on benchmark datasets. The framework demonstrates superior predictive ability for out-of-sample data and enhances the prediction of new reactions.

Rotational stability of monofocal and diffractive multifocal toric intraocular lens with identical design and material: a propensity score based prospective comparative study.

PURPOSE: To compare the rotational stability of a monofocal and a diffractive multifocal toric intraocular lens(IOLs) with identical design and material. METHODS: This prospective study enrolled patients who underwent plate-haptic toric IOL (AT TORBI 709 M and AT LISA 909 M) implantation. Propensity score matching (PSM) was performed to balance baseline factors. Follow-up examinations were conducted at 1 h, 1 day, 3 days, 1 week, 2 weeks, 1 month, and 3 months postoperatively. A linear mixed model of repeated measures was used to investigate the changes in IOL rotation over time. A 2-week timeframe was utilized to assess differences in IOL rotation between the two groups. RESULT: After PSM, a total of 126 eyes were selected from each group for further analysis. Postoperatively, the time course of IOL rotation change in the two groups remained consistent, with the greatest rotation occurring between 1 h and 1 day postoperatively. At the 2-week postoperative mark, the monofocal toric IOL exhibited a higher degree of rotation compared to the multifocal toric IOL (5.40 ± 7.77° vs. 3.53 ± 3.54°, P = 0.015). In lens thickness(LT) ≥ 4.5 mm and white-to-white distance(WTW) ≥ 11.6 mm subgroups, the monofocal toric IOL rotated greater than the multifocal toric IOL (P = 0.026 and P = 0.011, respectively). CONCLUSION: The diffractive multifocal toric IOL exhibits superior rotational stability compared to the monofocal toric IOL, especially in subgroups LT ≥ 4.5 mm and WTW ≥ 11.6 mm. Moreover, the time course of IOL rotation change is consistent for both, with the maximum rotation occurring between 1 h and 1 day postoperatively.

miR-1273h-5p protects the human corneal epithelium against UVR-induced oxidative stress and apoptosis: Role of miR-1273h-5p in climatic droplet keratopathy.

Climatic droplet keratopathy (CDK) is characterized by an increased number of oil-like deposits on the most anterior corneal layers, which affect vision and can cause blindness. Environmental ultraviolet radiation (UVR) exposure is a major risk factor, but the underlying mechanism of CDK pathogenesis is unclear. Increasing evidence has demonstrated that miRNAs participate in the cross-talk with oxidative stress. We aimed to explore whether certain miRNAs are involved in the pathogenesis of CDK. We performed miRNA sequencing of tears from patients with CDK and healthy individuals from Tacheng region of Xinjiang and conducted bioinformatic analysis of key miRNAs. We also evaluated viability, migration, and apoptosis of human corneal epithelial cells (HCECs) subjected to UVR treatment. miR-1273h-5p expression was abnormally downregulated in the tears of patients with CDK. miR-1273h-5p promoted cell proliferation and migration and inhibited UVR-induced mitochondrial apoptosis. miR-1273h-5p protected HCECs against UVR-induced oxidative damage by reducing the accumulation of reactive oxygen species and inhibiting mitochondrial apoptosis via the activation of the Nrf2 pathway. Thus, our results suggest that miR-1273h-5p protects the corneal epithelium against UVR-induced oxidative stress damage.

Inhibition of macrophages inflammasome activation via autophagic degradation of HMGB1 by EGCG ameliorates HBV-induced liver injury and fibrosis.

Background: Liver fibrosis is a reversible wound-healing response that can lead to end-stage liver diseases without effective treatment, in which HBV infection is a major cause. However, the underlying mechanisms for the development of HBV-induced fibrosis remains elusive, and efficacious therapies for this disease are still lacking. In present investigation, we investigated the effect and mechanism of green tea polyphenol epigallocatechin-3-gallate (EGCG) on HBV-induced liver injury and fibrosis. Methods: The effect of EGCG on liver fibrosis was examined in a recombinant cccDNA (rcccDNA) chronic HBV mouse model by immunohistochemical staining, Sirius red and Masson's trichrome staining. The functional relevance between high mobility group box 1 (HMGB1) and inflammasome activation and the role of EGCG in it were analyzed by Western blotting. The effect of EGCG on autophagic flux was determined by Western blotting and flow cytometric analysis. Results: EGCG treatment efficiently was found to alleviate HBV-induced liver injury and fibrosis in a recombinant cccDNA (rcccDNA) chronic HBV mouse model, a proven suitable research platform for HBV-induced fibrosis. Mechanistically, EGCG was revealed to repress the activation of macrophage NLRP3 inflammasome, a critical trigger of HBV-induced liver fibrosis. Further study revealed that EGCG suppressed macrophage inflammasome through downregulating the level of extracellular HMGB1. Furthermore, our data demonstrated that EGCG treatment downregulated the levels of extracellular HMGB1 through activating autophagic degradation of cytoplasmic HMGB1 in hepatocytes. Accordingly, autophagy blockade was revealed to significantly reverse EGCG-mediated inhibition on extracellular HMGB1-activated macrophage inflammasome and thus suppress the therapeutic effect of EGCG on HBV-induced liver injury and fibrosis. Conclusion: EGCG ameliorates HBV-induced liver injury and fibrosis via autophagic degradation of cytoplasmic HMGB1 and the subsequent suppression of macrophage inflammasome activation. These data provided a new pathogenic mechanism for HBV-induced liver fibrosis involving the extracellular HMGB1-mediated macrophage inflammasome activation, and also suggested EGCG administration as a promising therapeutic strategy for this disease.

ReLMole: Molecular Representation Learning Based on Two-Level Graph Similarities.

Molecular representation is a critical part of various prediction tasks for physicochemical properties of molecules and drug design. As graph notations are common in expressing the structural information of chemical compounds, graph neural networks (GNNs) have become the mainstream backbone model for learning molecular representation. However, the scarcity of task-specific labels in the biomedical domain limits the power of GNNs. Recently, self-supervised pretraining for GNNs has been leveraged to deal with this issue, while the existing pretraining methods are mainly designed for graph data in general domains without considering the specific data properties of molecules. In this paper, we propose a representation learning method for molecular graphs, called ReLMole, which is featured by a hierarchical graph modeling of molecules and a contrastive learning scheme based on two-level graph similarities. We assess the performance of ReLMole on two types of downstream tasks, namely, the prediction of molecular properties (MPs) and drug-drug interaction (DDIs). ReLMole achieves promising results for all the tasks. It outperforms the baseline models by over 2.6% on ROC-AUC averaged across six MP prediction tasks, and it improves the F1 value by 7-18% in DDI prediction for unseen drugs compared with other self-supervised models.

EmbedDTI: Enhancing the Molecular Representations via Sequence Embedding and Graph Convolutional Network for the Prediction of Drug-Target Interaction.

The identification of drug-target interaction (DTI) plays a key role in drug discovery and development. Benefitting from large-scale drug databases and verified DTI relationships, a lot of machine-learning methods have been developed to predict DTIs. However, due to the difficulty in extracting useful information from molecules, the performance of these methods is limited by the representation of drugs and target proteins. This study proposes a new model called EmbedDTI to enhance the representation of both drugs and target proteins, and improve the performance of DTI prediction. For protein sequences, we leverage language modeling for pretraining the feature embeddings of amino acids and feed them to a convolutional neural network model for further representation learning. For drugs, we build two levels of graphs to represent compound structural information, namely the atom graph and substructure graph, and adopt graph convolutional network with an attention module to learn the embedding vectors for the graphs. We compare EmbedDTI with the existing DTI predictors on two benchmark datasets. The experimental results show that EmbedDTI outperforms the state-of-the-art models, and the attention module can identify the components crucial for DTIs in compounds.

Evidence for the Involvement of Auxin, Ethylene and ROS Signaling During Primary Root Inhibition of Arabidopsis by the Allelochemical Benzoic Acid.

Allelopathy is mediated by plant-derived secondary metabolites (allelochemicals) which are released by donor plants and affect the growth and development of receptor plants. The plant root is the first organ which senses soil allelochemicals this results in the production of a shorter primary root. However, the mechanisms underlying this process remain elusive. Here, we report that a model allelochemical benzoic acid (BA) inhibited primary root elongation of Arabidopsis seedlings by reducing the sizes of both the meristem and elongation zones, and that auxin signaling affected this process. An increase in auxin level in the root tips was associated with increased expression of auxin biosynthesis genes and auxin polar transporter AUX1 and PIN2 genes under BA stress. Mutant analyses demonstrated that AUX1 and PIN2 rather than PIN1 were required for the inhibition of primary root elongation during BA exposure. Furthermore, BA stimulated ethylene evolution, whereas blocking BA-induced ethylene signaling with an ethylene biosynthesis inhibitor (Co2+), an ethylene perception antagonist (1-methylcyclopropene) or ethylene signaling mutant lines etr1-3 and ein3eil1 compromised BA-mediated inhibition of root elongation and up-regulation of auxin biosynthesis-related genes together with AUX1 and PIN2, indicating that ethylene signal was involved in auxin-mediated inhibition of primary root elongation during BA stress. Further analysis revealed that the BA-induced reactive oxygen species (ROS) burst contributed to BA-mediated root growth inhibition without affecting auxin and ethylene signals. Taken together, our results reveal that the allelochemical BA inhibits root elongation by increasing auxin accumulation via stimulation of auxin biosynthesis and AUX1/PIN2-mediated auxin transport via stimulation of ethylene production and an auxin/ethylene-independent ROS burst.

Auxin signalling of Arachis hypogaea activated by colonization of mutualistic fungus Phomopsis liquidambari enhances nodulation and N2 -fixation.

Beneficial fungal and rhizobial symbioses share commonalities in phytohormones responses, especially in auxin signalling. Mutualistic fungus Phomopsis liquidambari effectively increases symbiotic efficiency of legume peanut (Arachis hypogaea L.) with another microsymbiont, bradyrhizobium, but the underlying mechanisms are not well understood. We quantified and manipulated the IAA accumulation in ternary P. liquidambari-peanut-bradyrhizobial interactions to uncover its role between distinct symbioses. We found that auxin signalling is both locally and systemically induced by the colonization of P. liquidambari with peanut and further confirmed by Arabidopsis harbouring auxin-responsive reporter, DR5:GUS, and that auxin action, including auxin transport, is required to maintain fungal symbiotic behaviours and beneficial traits of plant during the symbiosis. Complementation and action inhibition experiments reveal that auxin signalling is involved in P. liquidambari-mediated nodule development and N2 -fixation enhancement and symbiotic gene activation. Further analyses showed that blocking of auxin action compromised the P. liquidambari-induced nodule phenotype and physiology changes, including vascular bundle development, symbiosome and bacteroids density, and malate concentrations, while induced the accumulation of starch granules in P. liquidambari-inoculated nodules. Collectively, our study demonstrated that auxin signalling activated by P. liquidambari symbiosis is recruited by peanut for bradyrhizobial symbiosis via symbiotic signalling pathway activation and nodule carbon metabolism enhancement.