Scientific discovery in the age of artificial intelligence.

Artificial intelligence (AI) is being increasingly integrated into scientific discovery to augment and accelerate research, helping scientists to generate hypotheses, design experiments, collect and interpret large datasets, and gain insights that might not have been possible using traditional scientific methods alone. Here we examine breakthroughs over the past decade that include self-supervised learning, which allows models to be trained on vast amounts of unlabelled data, and geometric deep learning, which leverages knowledge about the structure of scientific data to enhance model accuracy and efficiency. Generative AI methods can create designs, such as small-molecule drugs and proteins, by analysing diverse data modalities, including images and sequences. We discuss how these methods can help scientists throughout the scientific process and the central issues that remain despite such advances. Both developers and users of AI toolsneed a better understanding of when such approaches need improvement, and challenges posed by poor data quality and stewardship remain. These issues cut across scientific disciplines and require developing foundational algorithmic approaches that can contribute to scientific understanding or acquire it autonomously, making them critical areas of focus for AI innovation.

DSN-DDI: an accurate and generalized framework for drug-drug interaction prediction by dual-view representation learning.

Drug-drug interaction (DDI) prediction identifies interactions of drug combinations in which the adverse side effects caused by the physicochemical incompatibility have attracted much attention. Previous studies usually model drug information from single or dual views of the whole drug molecules but ignore the detailed interactions among atoms, which leads to incomplete and noisy information and limits the accuracy of DDI prediction. In this work, we propose a novel dual-view drug representation learning network for DDI prediction ('DSN-DDI'), which employs local and global representation learning modules iteratively and learns drug substructures from the single drug ('intra-view') and the drug pair ('inter-view') simultaneously. Comprehensive evaluations demonstrate that DSN-DDI significantly improved performance on DDI prediction for the existing drugs by achieving a relatively improved accuracy of 13.01% and an over 99% accuracy under the transductive setting. More importantly, DSN-DDI achieves a relatively improved accuracy of 7.07% to unseen drugs and shows the usefulness for real-world DDI applications. Finally, DSN-DDI exhibits good transferability on synergistic drug combination prediction and thus can serve as a generalized framework in the drug discovery field.

Breaking the barriers of data scarcity in drug-target affinity prediction.

Accurate prediction of drug-target affinity (DTA) is of vital importance in early-stage drug discovery, facilitating the identification of drugs that can effectively interact with specific targets and regulate their activities. While wet experiments remain the most reliable method, they are time-consuming and resource-intensive, resulting in limited data availability that poses challenges for deep learning approaches. Existing methods have primarily focused on developing techniques based on the available DTA data, without adequately addressing the data scarcity issue. To overcome this challenge, we present the Semi-Supervised Multi-task training (SSM) framework for DTA prediction, which incorporates three simple yet highly effective strategies: (1) A multi-task training approach that combines DTA prediction with masked language modeling using paired drug-target data. (2) A semi-supervised training method that leverages large-scale unpaired molecules and proteins to enhance drug and target representations. This approach differs from previous methods that only employed molecules or proteins in pre-training. (3) The integration of a lightweight cross-attention module to improve the interaction between drugs and targets, further enhancing prediction accuracy. Through extensive experiments on benchmark datasets such as BindingDB, DAVIS and KIBA, we demonstrate the superior performance of our framework. Additionally, we conduct case studies on specific drug-target binding activities, virtual screening experiments, drug feature visualizations and real-world applications, all of which showcase the significant potential of our work. In conclusion, our proposed SSM-DTA framework addresses the data limitation challenge in DTA prediction and yields promising results, paving the way for more efficient and accurate drug discovery processes.

R2-DDI: relation-aware feature refinement for drug-drug interaction prediction.

Precisely predicting the drug-drug interaction (DDI) is an important application and host research topic in drug discovery, especially for avoiding the adverse effect when using drug combination treatment for patients. Nowadays, machine learning and deep learning methods have achieved great success in DDI prediction. However, we notice that most of the works ignore the importance of the relation type when building the DDI prediction models. In this work, we propose a novel R$^2$-DDI framework, which introduces a relation-aware feature refinement module for drug representation learning. The relation feature is integrated into drug representation and refined in the framework. With the refinement features, we also incorporate the consistency training method to regularize the multi-branch predictions for better generalization. Through extensive experiments and studies, we demonstrate our R$^2$-DDI approach can significantly improve the DDI prediction performance over multiple real-world datasets and settings, and our method shows better generalization ability with the help of the feature refinement design.

Functional Composite Dual-Phase In Situ Self-Reconstruction Design for High-Energy-Density Li-Rich Cathodes.

The unique anionic redox mechanism provides, high-capacity, irreversible oxygen release and voltage/capacity degradation to Li-rich cathode materials (LRO, Li1.2Mn0.54Co0.13Ni0.13O2). In this study, an integrated stabilized carbon-rock salt/spinel composite heterostructured layers (C@spinel/MO) is constructed by in situ self-reconstruction, and the generation mechanism of the in situ reconstructed surface is elucidated. The formation of atomic-level connections between the surface-protected phase and bulk-layered phase contributes to electrochemical performance. The best-performing sample shows a high increase (63%) of capacity retention compared to that of the pristine sample after 100 cycles at 1C, with an 86.7% reduction in surface oxygen release shown by differential electrochemical mass spectrometry. Soft X-ray results show that Co3+ and Mn4+ are mainly reduce in the carbothermal reduction reaction and participate in the formation of the spinel/MO rock-salt phase. The results of oxygen release characterized by Differential electrochemical mass spectrometry (DEMS) strongly prove the effectiveness of surface reconstruction.

BioGPT: generative pre-trained transformer for biomedical text generation and mining.

Pre-trained language models have attracted increasing attention in the biomedical domain, inspired by their great success in the general natural language domain. Among the two main branches of pre-trained language models in the general language domain, i.e. BERT (and its variants) and GPT (and its variants), the first one has been extensively studied in the biomedical domain, such as BioBERT and PubMedBERT. While they have achieved great success on a variety of discriminative downstream biomedical tasks, the lack of generation ability constrains their application scope. In this paper, we propose BioGPT, a domain-specific generative Transformer language model pre-trained on large-scale biomedical literature. We evaluate BioGPT on six biomedical natural language processing tasks and demonstrate that our model outperforms previous models on most tasks. Especially, we get 44.98%, 38.42% and 40.76% F1 score on BC5CDR, KD-DTI and DDI end-to-end relation extraction tasks, respectively, and 78.2% accuracy on PubMedQA, creating a new record. Our case study on text generation further demonstrates the advantage of BioGPT on biomedical literature to generate fluent descriptions for biomedical terms.

Improved drug-target interaction prediction with intermolecular graph transformer.

The identification of active binding drugs for target proteins (referred to as drug-target interaction prediction) is the key challenge in virtual screening, which plays an essential role in drug discovery. Although recent deep learning-based approaches achieve better performance than molecular docking, existing models often neglect topological or spatial of intermolecular information, hindering prediction performance. We recognize this problem and propose a novel approach called the Intermolecular Graph Transformer (IGT) that employs a dedicated attention mechanism to model intermolecular information with a three-way Transformer-based architecture. IGT outperforms state-of-the-art (SoTA) approaches by 9.1% and 20.5% over the second best option for binding activity and binding pose prediction, respectively, and exhibits superior generalization ability to unseen receptor proteins than SoTA approaches. Furthermore, IGT exhibits promising drug screening ability against severe acute respiratory syndrome coronavirus 2 by identifying 83.1% active drugs that have been validated by wet-lab experiments with near-native predicted binding poses. Source code and datasets are available at https://github.com/microsoft/IGT-Intermolecular-Graph-Transformer.

SPRoBERTa: protein embedding learning with local fragment modeling.

Well understanding protein function and structure in computational biology helps in the understanding of human beings. To face the limited proteins that are annotated structurally and functionally, the scientific community embraces the self-supervised pre-training methods from large amounts of unlabeled protein sequences for protein embedding learning. However, the protein is usually represented by individual amino acids with limited vocabulary size (e.g. 20 type proteins), without considering the strong local semantics existing in protein sequences. In this work, we propose a novel pre-training modeling approach SPRoBERTa. We first present an unsupervised protein tokenizer to learn protein representations with local fragment pattern. Then, a novel framework for deep pre-training model is introduced to learn protein embeddings. After pre-training, our method can be easily fine-tuned for different protein tasks, including amino acid-level prediction task (e.g. secondary structure prediction), amino acid pair-level prediction task (e.g. contact prediction) and also protein-level prediction task (remote homology prediction, protein function prediction). Experiments show that our approach achieves significant improvements in all tasks and outperforms the previous methods. We also provide detailed ablation studies and analysis for our protein tokenizer and training framework.

Tryptophan catabolism via the kynurenine pathway regulates infection and inflammation: from mechanisms to biomarkers and therapies.

BACKGROUND: L-Tryptophan (L-Trp), an essential amino acid, is the only amino acid whose level is regulated specifically by immune signals. Most proportions of Trp are catabolized via the kynurenine (Kyn) pathway (KP) which has evolved to align the food availability and environmental stimulation with the host pathophysiology and behavior. Especially, the KP plays an indispensable role in balancing the immune activation and tolerance in response to pathogens. SCOPE OF REVIEW: In this review, we elucidate the underlying immunological regulatory network of Trp and its KP-dependent catabolites in the pathophysiological conditions by participating in multiple signaling pathways. Furthermore, the KP-based regulatory roles, biomarkers, and therapeutic strategies in pathologically immune disorders are summarized covering from acute to chronic infection and inflammation. MAJOR CONCLUSIONS: The immunosuppressive effects dominate the functions of KP induced-Trp depletion and KP-produced metabolites during infection and inflammation. However, the extending minor branches from the KP are not confined to the immune tolerance, instead they go forward to various functions according to the specific condition. Nevertheless, persistent efforts should be made before the clinical use of KP-based strategies to monitor and cure infectious and inflammatory diseases.

A Case of Invasive Pulmonary Mycosis (Rhizopus microspores) secondary to Hematological Disease.

BACKGROUND: Pulmonary mucormycosis is most common in patients with hematologic malignancies and transplant recipients. This article describes a case of mucormycosis in the lungs secondary to a hematologic disorder with suspected lung cancer. METHODS: Rhizopus (Rhizopus microspores) was detected by blood NGS and bronchoalveolar lavage fluid NGS, and pulmonary mucormycosis was confirmed. RESULTS: Secondary to hematologic disease, pulmonary pneumonia, mycosis, and symptoms improved after comprehensive treatment. CONCLUSIONS: Clinical data and radiologic knowledge are combined to diagnose invasive pulmonary mycoses; early empirical medicine is very important.

Multiple Postoperative Lung Infections after Thymoma Surgery Diagnosed as Nontuberculous Mycobacterial Infection.

BACKGROUND: Thymomas are thymic epithelial-derived, most common primary anterior mediastinal masses. Non-tuberculous mycobacteria (NTM) are species that do not cause leprosy and belong to species outside the Mycobacterium tuberculosis complex. METHODS: With the clinical application of targeted next-generation sequencing (tNGS), we promptly confirmed a case of NTM infection combined with NTM infection after thymoma surgery, and we performed a joint literature analysis of the two diseases to improve clinicians' understanding and recognition of lung infections after thymoma surgery. RESULTS: Chest CT of both lungs showed multiple hyperdense shadows. Sputum bacterial culture and characterization detected Neisseria Dryad and Streptococcus Grass Green. The presence of Mycobacterium abscessus infection was confirmed by alveolar lavage fluid sent for second-generation macro gene sequencing. CONCLUSIONS: The body's immune function decreases after thymoma surgery. When empirical anti-infection treatment for recurrent pneumonia in the lungs is ineffective, we should be alerted to the possibility of the presence of pulmonary non-tuberculous mycobacterial infection, and next-generation sequencing should be performed promptly to arrive quickly at a diagnosis.

Targeted inhibition of the PI3K/AKT/mTOR pathway by (+)-anthrabenzoxocinone induces cell cycle arrest, apoptosis, and autophagy in non-small cell lung cancer.

Non-small cell lung cancer (NSCLC), characterized by low survival rates and a high recurrence rate, is a major cause of cancer-related mortality. Aberrant activation of the PI3K/AKT/mTOR signaling pathway is a common driver of NSCLC. Within this study, the inhibitory activity of (+)-anthrabenzoxocinone ((+)-ABX), an oxygenated anthrabenzoxocinone compound derived from Streptomyces, against NSCLC is demonstrated for the first time both in vitro and in vivo. Mechanistically, it is confirmed that the PI3K/AKT/mTOR signaling pathway is targeted and suppressed by (+)-ABX, resulting in the induction of S and G2/M phase arrest, apoptosis, and autophagy in NSCLC cells. Additionally, the augmentation of intracellular ROS levels by (+)-ABX is revealed, further contributing to the inhibition of the signaling pathway and exerting inhibitory effects on tumor growth. The findings presented in this study suggest that (+)-ABX possesses the potential to serve as a lead compound for the treatment of NSCLC.

Mesenchymal stem cell-derived exosomes as a new drug carrier for the treatment of spinal cord injury: A review.

Spinal cord injury (SCI) is a devastating traumatic disease seriously impairing the quality of life in patients. Expectations to allow the hopeless central nervous system to repair itself after injury are unfeasible. Developing new approaches to regenerate the central nervous system is still the priority. Exosomes derived from mesenchymal stem cells (MSC-Exo) have been proven to robustly quench the inflammatory response or oxidative stress and curb neuronal apoptosis and autophagy following SCI, which are the key processes to rescue damaged spinal cord neurons and restore their functions. Nonetheless, MSC-Exo in SCI received scant attention. In this review, we reviewed our previous work and other studies to summarize the roles of MSC-Exo in SCI and its underlying mechanisms. Furthermore, we also focus on the application of exosomes as drug carrier in SCI. In particular, it combs the advantages of exosomes as a drug carrier for SCI, imaging advantages, drug types, loading methods, etc., which provides the latest progress for exosomes in the treatment of SCI, especially drug carrier.

Neoerysiphe leontopodii sp. nov., a new powdery mildew from China.

A powdery mildew was found on Leontopodium leontopodioides (Asteraceae) in China. Phylogenetic analyses using a combination of internal transcribed spacer and 28S rDNA sequences showed that this species, which clusters as sister to Neoerysiphe joerstadii, is allied to N. galii, N. geranii, and N. nevoi. This species differs from the closely allied N. joerstadii in the number and size of asci (3-10 asci, 55-75 × 20-40 µm versus 16-32 asci, 40-60 × 20-30 µm). This species is morphologically very similar to N. gnaphalii, but clearly differs from this species in having larger chasmothecia and colorless appendages. Therefore, the powdery mildew on L. leontopodioides is described as N. leontopodii sp. nov.

[Construction of a risk prediction model for bronchiolitis obliterans in children with refractory Mycoplasma pneumoniae pneumonia]. / 难治性肺炎支原体肺炎患儿并发闭塞性细支气管炎风险预测模型的构建.

OBJECTIVES: To explore the establishment of a risk prediction model for concurrent bronchiolitis obliterans (BO) in children with refractory Mycoplasma pneumoniae pneumonia (RMPP). METHODS: A retrospective study included 116 RMPP children treated in the Department of Pediatrics of Xiangya Changde Hospital from June 2021 to December 2023. Eighty-one cases were allocated to the training set and thirty-five cases to the validation set based on a 7:3 ratio. Among them, 26 cases in the training set developed BO, while 55 did not. The multivariate logistic regression was used to select variable factors for constructing the BO risk prediction model. Nomograms were drawn, and the receiver operating characteristic (ROC) curve was used to assess the discriminative ability of the model, while calibration curves and Hosmer-Lemeshow tests evaluated the model's calibration. RESULTS: Multivariate logistic regression analysis indicated that several factors were significantly associated with concurrent BO in RMPP children, including length of hospital stay, duration of fever, atelectasis, neutrophil percentage (NEUT%), peak lactate dehydrogenase (LDH), ferritin, peak C reactive protein (CRP), oxygenation index (PaO2/FiO2), ≥2/3 lung lobe consolidation, pleural effusion, bronchial mucous plugs, bronchial mucosal necrosis, and arterial oxygen partial pressure (PaO2) (P<0.05). ROC curve analysis for the training set indicated an area under the curve of 0.904 with 88% sensitivity and 83% specificity; the validation set showed an area under the curve of 0.823 with 76% sensitivity and 93% specificity. The Hosmer-Lemeshow test's Chi-square values for the training and validation sets were 2.17 and 1.92, respectively, with P values of 0.221 and 0.196, respectively. CONCLUSIONS: The risk prediction model for BO in RMPP children based on logistic regression has good performance. Variables such as length of hospital stay, duration of fever, atelectasis, peak LDH, peak CRP, NEUT%, ferritin, ≥2/3 lung lobe consolidation, pleural effusion, bronchial mucous plugs, bronchial mucosal necrosis, PaO2/FiO2, andPaO2 can be used as predictors.

Discovering drug-target interaction knowledge from biomedical literature.

MOTIVATION: The interaction between drugs and targets (DTI) in human body plays a crucial role in biomedical science and applications. As millions of papers come out every year in the biomedical domain, automatically discovering DTI knowledge from biomedical literature, which are usually triplets about drugs, targets and their interaction, becomes an urgent demand in the industry. Existing methods of discovering biological knowledge are mainly extractive approaches that often require detailed annotations (e.g. all mentions of biological entities, relations between every two entity mentions, etc.). However, it is difficult and costly to obtain sufficient annotations due to the requirement of expert knowledge from biomedical domains. RESULTS: To overcome these difficulties, we explore an end-to-end solution for this task by using generative approaches. We regard the DTI triplets as a sequence and use a Transformer-based model to directly generate them without using the detailed annotations of entities and relations. Further, we propose a semi-supervised method, which leverages the aforementioned end-to-end model to filter unlabeled literature and label them. Experimental results show that our method significantly outperforms extractive baselines on DTI discovery. We also create a dataset, KD-DTI, to advance this task and release it to the community. AVAILABILITY AND IMPLEMENTATION: Our code and data are available at https://github.com/bert-nmt/BERT-DTI. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Improving machine learning force fields for molecular dynamics simulations with fine-grained force metrics.

Machine learning force fields (MLFFs) have gained popularity in recent years as they provide a cost-effective alternative to ab initio molecular dynamics (MD) simulations. Despite a small error on the test set, MLFFs inherently suffer from generalization and robustness issues during MD simulations. To alleviate these issues, we propose global force metrics and fine-grained metrics from element and conformation aspects to systematically measure MLFFs for every atom and every conformation of molecules. We selected three state-of-the-art MLFFs (ET, NequIP, and ViSNet) and comprehensively evaluated on aspirin, Ac-Ala3-NHMe, and Chignolin MD datasets with the number of atoms ranging from 21 to 166. Driven by the trained MLFFs on these molecules, we performed MD simulations from different initial conformations, analyzed the relationship between the force metrics and the stability of simulation trajectories, and investigated the reason for collapsed simulations. Finally, the performance of MLFFs and the stability of MD simulations can be further improved guided by the proposed force metrics for model training, specifically training MLFF models with these force metrics as loss functions, fine-tuning by reweighting samples in the original dataset, and continued training by recruiting additional unexplored data.

Risk factors associated with loss of variable-thread tapered implants: A retrospective observational study of 1-5 years.

OBJECTIVES: This study aimed to evaluate the survival rate of variable-thread tapered implants (VTTIs) and identify risk factors for early/late implant loss. MATERIALS AND METHODS: From January 2016 to December 2019, patients who received VTTIs were included in this study. The cumulative survival rates (CSRs) at implant/patient levels were calculated by the life table method and presented via Kaplan-Meier survival curves. The relation between investigated variables and early/late implant loss was analyzed by the multivariate generalized estimating equation (GEE) regression model on the implant level. RESULTS: A total of 1528 patients with 2998 VTTIs were included. 95 implants from 76 patients were lost at the end of observation. At the implant level, the CSRs at 1, 3, and 5 years were 98.77%, 96.97%, and 95.39%, respectively, whereas they were 97.84%, 95.31%, and 92.96% at the patient level, respectively. The multivariate analysis revealed that non-submerged implant healing (OR = 4.63, p = .037) was associated with the early loss of VTTIs. Besides, male gender (OR = 2.48, p = .002), periodontitis (OR = 3.25, p = .007), implant length <10 mm (OR = 2.63, p = .028), and overdenture (OR = 9.30, p = .004) could significantly increase the risk of late implant loss. CONCLUSION: Variable-thread tapered implants could reach an acceptable survival rate in clinical practice. Non-submerged implant healing was associated with early implant loss; male gender, periodontitis, implant length <10 mm, and overdenture would significantly increase the risk of late implant loss.

Indicator-based assessments of the coupling coordination degree and correlations of water-energy-food-ecology nexus in Uzbekistan.

Grappling with the global ecological concern of the Aral Sea disaster, Uzbekistan exemplifies the urgent necessity of unravelling and addressing the complex Water-Energy-Food-Ecology (WEFE) nexus conflicts in arid regions, a critical task yet largely uncharted. Through the strategic process of 'Indicator Articulation - Weight Calibration - Nexus Coordination Quantification - Correlational Analysis', this work has developed a tailored framework that integrates a novel, context-specific indicator system, enabling an illumination of the intricate dynamics within the WEFE nexus in arid regions. During 2000-2018, the WEFE Nexus in Uzbekistan showed low-level coordination, indicating systemic imbalances. The Aral Sea crisis was the central disruptor, resulting in a moderately disordered ecological subsystem. Concurrently, disorder was observed in water resources, signaling inadequate management and potential overutilization. Furthermore, Coordination for energy and food were barely coordinated and under primary coordination respectively, underlining critical challenges in energy efficiency and food security. Over the last two decades, the WEFE Nexus has evolved towards a tighter interlinkage, yet the stability of this coupling coordination has experienced increased fluctuations, indicating that Uzbekistan's policies in the WEFE subsystems have been less stable in the last two decades and are in need of further adjustment and improvement. To address the challenges, we recommend a comprehensive approach that integrates technological, infrastructure, and policy solutions is needed. Specifically, promoting water-saving irrigation technology, renewing and maintaining outdated energy facilities, and raising public awareness of ecological protection are part of the essential measures. Furthermore, alleviating the contradiction between economic growth and ecological conservation remains a major challenge. Collectively, our constructed WEFE Nexus framework, with its extendable and context-specific indicators, holds significant potential for broad application in the analysis of multi-sectoral sustainability, particularly within arid regions globally, and forms a solid foundation for the formulation of effective, targeted policies and sustainable development strategies.