RESUMO
Prediction of drug-target interactions (DTIs) is essential in medicine field, since it benefits the identification of molecular structures potentially interacting with drugs and facilitates the discovery and reposition of drugs. Recently, much attention has been attracted to network representation learning to learn rich information from heterogeneous data. Although network representation learning algorithms have achieved success in predicting DTI, several manually designed meta-graphs limit the capability of extracting complex semantic information. To address the problem, we introduce an adaptive meta-graph-based method, termed AMGDTI, for DTI prediction. In the proposed AMGDTI, the semantic information is automatically aggregated from a heterogeneous network by training an adaptive meta-graph, thereby achieving efficient information integration without requiring domain knowledge. The effectiveness of the proposed AMGDTI is verified on two benchmark datasets. Experimental results demonstrate that the AMGDTI method overall outperforms eight state-of-the-art methods in predicting DTI and achieves the accurate identification of novel DTIs. It is also verified that the adaptive meta-graph exhibits flexibility and effectively captures complex fine-grained semantic information, enabling the learning of intricate heterogeneous network topology and the inference of potential drug-target relationship.
Assuntos
Algoritmos , Medicina , Benchmarking , Sistemas de Liberação de Medicamentos , SemânticaRESUMO
BACKGROUND: It's unclear if excess visceral adipose tissue (VAT) mass in individuals with prediabetes can be countered by adherence to a Mediterranean lifestyle (MEDLIFE). We aimed to examine VAT mass, MEDLIFE adherence, and their impact on type 2 diabetes (T2D) and diabetic microvascular complications (DMC) in individuals with prediabetes. METHODS: 11,267 individuals with prediabetes from the UK Biobank cohort were included. VAT mass was predicted using a non-linear model, and adherence to the MEDLIFE was evaluated using the 25-item MEDLIFE index, encompassing categories such as "Mediterranean food consumption," "Mediterranean dietary habits," and "Physical activity, rest, social habits, and conviviality." Both VAT and MEDLIFE were categorized into quartiles, resulting in 16 combinations. Incident cases of T2D and related DMC were identified through clinical records. Cox proportional-hazards regression models were employed to examine associations, adjusting for potential confounding factors. RESULTS: Over a median follow-up of 13.77 years, we observed 1408 incident cases of T2D and 714 cases of any DMC. High adherence to the MEDLIFE, compared to the lowest quartile, reduced a 16% risk of incident T2D (HR: 0.84, 95% CI: 0.71-0.98) and 31% for incident DMC (0.69, 0.56-0.86). Conversely, compared to the lowest quartile of VAT, the highest quartile increased the risk of T2D (5.95, 4.72-7.49) and incident any DMC (1.79, 1.36-2.35). We observed an inverse dose-response relationship between MEDLIFE and T2D/DMC, and a dose-response relationship between VAT and all outcomes (P for trend < 0.05). Restricted cubic spline analysis confirmed a nearly linear dose-response pattern across all associations. Compared to individuals with the lowest MEDLIFE quartile and highest VAT quartile, those with the lowest T2D risk had the lowest VAT and highest MEDLIFE (0.12, 0.08-0.19). High MEDLIFE was linked to reduced T2D risk across all VAT categories, except in those with the highest VAT quartile. Similar trends were seen for DMC. CONCLUSION: High adherence to MEDLIFE reduced T2D and MDC risk in individuals with prediabetes, while high VAT mass increases it, but MEDLIFE adherence may offset VAT's risk partly. The Mediterranean lifestyle's adaptability to diverse populations suggests promise for preventing T2D.
Assuntos
Diabetes Mellitus Tipo 2 , Angiopatias Diabéticas , Dieta Mediterrânea , Gordura Intra-Abdominal , Estado Pré-Diabético , Fatores de Proteção , Comportamento de Redução do Risco , Humanos , Estado Pré-Diabético/epidemiologia , Estado Pré-Diabético/diagnóstico , Diabetes Mellitus Tipo 2/diagnóstico , Diabetes Mellitus Tipo 2/epidemiologia , Masculino , Feminino , Pessoa de Meia-Idade , Gordura Intra-Abdominal/fisiopatologia , Idoso , Fatores de Risco , Medição de Risco , Angiopatias Diabéticas/epidemiologia , Angiopatias Diabéticas/diagnóstico , Angiopatias Diabéticas/prevenção & controle , Fatores de Tempo , Incidência , Adiposidade , Reino Unido/epidemiologia , Adulto , Dieta Saudável , Exercício Físico , Estilo de Vida Saudável , Obesidade Abdominal/diagnóstico , Obesidade Abdominal/epidemiologia , Obesidade Abdominal/fisiopatologia , Estudos ProspectivosRESUMO
Immune checkpoint therapy (ICT) has greatly improved the survival of cancer patients in the past few years, but only a small number of patients respond to ICT. To predict ICT response, we developed a multi-modal feature fusion model based on deep learning (MFMDL). This model utilizes graph neural networks to map gene-gene relationships in gene networks to low dimensional vector spaces, and then fuses biological pathway features and immune cell infiltration features to make robust predictions of ICT. We used five datasets to validate the predictive performance of the MFMDL. These five datasets span multiple types of cancer, including melanoma, lung cancer, and gastric cancer. We found that the prediction performance of multi-modal feature fusion model based on deep learning is superior to other traditional ICT biomarkers, such as ICT targets or tumor microenvironment-associated markers. In addition, we also conducted ablation experiments to demonstrate the necessity of fusing different modal features, which can improve the prediction accuracy of the model.
Assuntos
Aprendizado Profundo , Neoplasias Pulmonares , Melanoma , Humanos , Imunoterapia , Redes Reguladoras de Genes , Neoplasias Pulmonares/terapia , Microambiente TumoralRESUMO
This article studies the generalization of neural networks (NNs) by examining how a network changes when trained on a training sample with or without out-of-distribution (OoD) examples. If the network's predictions are less influenced by fitting OoD examples, then the network learns attentively from the clean training set. A new notion, dataset-distraction stability, is proposed to measure the influence. Extensive CIFAR-10/100 experiments on the different VGG, ResNet, WideResNet, ViT architectures, and optimizers show a negative correlation between the dataset-distraction stability and generalizability. With the distraction stability, we decompose the learning process on the training set S into multiple learning processes on the subsets of S drawn from simpler distributions, i.e., distributions of smaller intrinsic dimensions (IDs), and furthermore, a tighter generalization bound is derived. Through attentive learning, miraculous generalization in deep learning can be explained and novel algorithms can also be designed.
RESUMO
The prediction of molecular properties remains a challenging task in the field of drug design and development. Recently, there has been a growing interest in the analysis of biological images. Molecular images, as a novel representation, have proven to be competitive, yet they lack explicit information and detailed semantic richness. Conversely, semantic information in SMILES sequences is explicit but lacks spatial structural details. Therefore, in this study, we focus on and explore the relationship between these two types of representations, proposing a novel multimodal architecture named ISMol. ISMol relies on a cross-attention mechanism to extract information representations of molecules from both images and SMILES strings, thereby predicting molecular properties. Evaluation results on 14 small molecule ADMET datasets indicate that ISMol outperforms machine learning (ML) and deep learning (DL) models based on single-modal representations. In addition, we analyze our method through a large number of experiments to test the superiority, interpretability and generalizability of the method. In summary, ISMol offers a powerful deep learning toolbox for drug discovery in a variety of molecular properties.
Assuntos
Desenho de Fármacos , Descoberta de Drogas , Humanos , Aprendizado de Máquina , SemânticaRESUMO
Corynebacterium glutamicum was used to ferment wheat gluten hydrolysates (WGHs) to prepare flavour base. This study investigated the effect of hydrolysis degrees (DHs) and fermentation time on flavour of WGHs. During fermentation, the contents of amino nitrogen, total acid and small peptides increased, while the protein and pH value decreased. Succinic acid, GMP, and Glu were the prominent umami substances in fermented WGHs. The aromas of WGHs with different DHs could be distinguished by electronic nose and GC-IMS. Based on OAV of GC-MS, hexanal was the main compound in WGHs, while phenylethyl alcohol and acetoin were dominant after fermentation. WGHs with high DHs accumulated more flavour metabolites. Correlation analysis showed that small peptides (<1 kDa) could promote the formation of flavour substances, and Asp was potentially relevant flavour precursor. This study indicated that fermented WGHs with different DHs can potentially be used in different food applications based on flavour profiles.
RESUMO
BACKGROUND: Healthier lifestyle decreased the risk of mental disorders (MDs) such as depression and anxiety. However, research on the effects of a comprehensive healthy lifestyle on their progression is lacking. METHODS: 385,704 individuals without baseline MDs from the UK Biobank cohort were included. A composite healthy lifestyle score was computed by assessing alcohol intake, smoking status, television viewing time, physical activity, sleep duration, fruit and vegetable intake, oily fish intake, red meat intake, and processed meat intake. Follow-up utilized hospital and death register records. Multistate model was used to examine the role of healthy lifestyle on the progression of specific MDs, while a piecewise Cox regression model was utilized to assess the influence of healthy lifestyle across various phases of disease progression. RESULTS: Higher lifestyle score reduced risks of transitions from baseline to anxiety and depression, as well as from anxiety and depression to comorbidity, with corresponding hazard ratios (HR) and 95 % confidence intervals (CI) of 0.94 (0.93, 0.95), 0.90 (0.89, 0.91), 0.94 (0.91, 0.98), and 0.95 (0.92, 0.98), respectively. Healthier lifestyle decreased the risk of transitioning from anxiety to comorbidity within 2 years post-diagnosis, with HR 0.93 (0.88, 0.98). Higher lifestyle scores at 2-4 years and 4-6 years post-depression onset were associated with reduced risk of comorbidity, with HR 0.93 (0.87, 0.99) and 0.92 (0.86, 0.99), respectively. LIMITATION: The generalizability to other ethnic groups is limited. CONCLUSION: This study observed a protective role of holistic healthy lifestyle in the trajectory of MDs and contributed to identifying critical progression windows.