Independent Datasets Research Articles

Causal relationship between gut microbiota and lung squamous cell carcinoma: a bidirectional two-sample Mendelian randomization study.

This study aims to explore the potential causal relationship between gut microbiota and lung squamous cell carcinoma (LUSC). A bidirectional two-sample Mendelian randomization analysis was conducted using genome-wide association study (GWAS) data from gut microbiota and LUSC. Gut microbiota served as the exposure factor, with instrumental variables selected from a GWAS involving 18 340 participants. LUSC data were drawn from a European cohort including 29 266 LUSC cases and 56 450 controls. Inverse-variance weighted (IVW) method was used as the primary method, with the Benjamini-Hochberg method applied to adjust for multiple comparisons. An independent dataset (ieu-a-967, containing 3275 LUSC cases and 15 038 controls) was used for replication analysis to ensure robustness. IVW analysis found that Butyricicoccus (OR = 0.79, 95% CI: 0.63-0.99, P = .042) and Coprobacter (OR = 0.85, 95% CI: 0.74-0.97, P = .018) were significantly protective against LUSC. In contrast, Victivallis (OR = 1.11, 95% CI: 1.00-1.23, P = .045) and Ruminococcus (OR = 1.28, 95% CI: 1.03-1.60, P = .028) increased LUSC risk. Replication analysis in the independent dataset confirmed significant associations for Ruminococcus and Coprobacter. No reverse causality or pleiotropy was detected. This study provides evidence of a causal relationship between specific gut microbiota and LUSC risk, highlighting new microbial targets for potential prevention and treatment strategies in lung cancer. Key messages What is already known on this topic? Previous studies have suggested potential links between gut microbiota composition and the development of various cancers, including lung cancer. However, the exact causal relationship between specific gut microbiota and lung squamous cell carcinoma (LUSC) has remained unclear. Traditional observational studies have struggled to determine the direction of causality due to confounding factors, making further investigation necessary through more robust methods such as Mendelian randomization (MR). What this study adds? This bidirectional MR study provides novel genetic evidence indicating that certain gut microbiotas are causally associated with LUSC risk. Specifically, Butyricicoccus appears to reduce the risk of LUSC, while Victivallis increases the risk. These findings highlight the role of the gut-lung axis in LUSC and open up new avenues for exploring gut microbiota as potential modulators of lung cancer risk. How this study might affect research, practice, or policy? The implications of this study may significantly influence future research into cancer prevention strategies by targeting gut microbiota. Additionally, it could inform clinical practices aimed at modulating gut microbiota to lower the risk of LUSC, potentially influencing dietary or probiotic interventions to reduce cancer susceptibility. Furthermore, these results might shape public health policies that focus on the gut-lung axis as a novel avenue for cancer prevention and management.

Unraveling the causal relationship between serum minerals and pancreatic cancer: a Mendelian randomization study

BackgroundPancreatic cancer is among the most lethal malignancies, characterized by a poor prognosis and limited modifiable factors. Emerging evidence indicates that serum mineral levels may influence the likelihood of developing pancreatic cancer. However, the causal relationship between serum minerals and pancreatic cancer remains unclear and warrants further investigation.MethodsThis Mendelian randomization (MR) study was conducted to explore the causal effects of serum mineral levels on pancreatic cancer risk. Genetic variants associated with serum mineral levels, including calcium, iron, magnesium, zinc, selenium, and copper, were selected as instrumental variables (IVs) from large-scale genome-wide association study (GWAS) data. Multiple methods, including inverse variance weighting (IVW), MR-Egger, weighted median, weight methods, were employed to perform MR analysis. The effect sizes from the MR analysis, using two independent GWAS summary datasets related to pancreatic cancer, were combined through meta-analysis. The Cochrane Q test, MR-Egger intercept test, MR-PRESSO global test, and leave-one-out test were conducted for sensitivity tests.ResultsOur MR analysis demonstrated a significant causal effect of genetically predicted serum calcium levels on increased pancreatic cancer risk [OR = 1.64, 95% CI 1.05–2.57, P = 0.029 (discovery cohort); OR = 1.52, 95% CI 1.07–2.15, P = 0.019 (validation cohort)], while no significant associations were found for other serum minerals (P > 0.05). Additional meta-analysis reinforces and substantiates this conclusion (pooled OR = 1.56, 95% CI 1.19–2.06, P = 0.001). No evidence of pleiotropy or heterogeneity was detected across multiple sensitivity tests (P > 0.05).ConclusionThis study provides new evidence supporting the causal role of certain serum minerals, particularly calcium, in the development of pancreatic cancer. These findings may help inform future research into preventive strategies or therapies aimed at modulating mineral levels in patients at high risk of pancreatic cancer.

Monitoring of the trough concentration of valproic acid in pediatric epilepsy patients: a machine learning-based ensemble model

AimsFew personalized monitoring models for valproic acid (VPA) in pediatric epilepsy patients (PEPs) incorporate machine learning (ML) algorithms. This study aimed to develop an ensemble ML model for VPA monitoring to enhance clinical precision of VPA usage.MethodsA dataset comprising 366 VPA trough concentrations from 252 PEPs, along with 19 covariates and the target variable (VPA trough concentration), was refined by Spearman correlation and multicollinearity testing (366 × 11). The dataset was split into a training set (292) and testing set (74) at a ratio of 8:2. An ensemble model was formulated by Gradient Boosting Regression Trees (GBRT), Random Forest Regression (RFR), and Support Vector Regression (SVR), and assessed by SHapley Additive exPlanations (SHAP) analysis for covariate importance. The model was optimized for R2, relative accuracy, and absolute accuracy, and validated against two independent external datasets (32 in-hospital and 28 out-of-hospital dataset).ResultsUsing the R2 weight ratio of GBRT, RFR and SVR optimized at 5:2:3, the ensemble model demonstrated superior performance in terms of relative accuracy (87.8%), absolute accuracy (78.4%), and R2 (0.50), while also exhibiting a lower Mean Absolute Error (9.87) and Root Mean Squared Error (12.24), as validated by the external datasets. Platelet count (PLT) and VPA daily dose were identified as pivotal covariates.ConclusionThe proposed ensemble model effectively monitors VPA trough concentrations in PEPs. By integrating covariates across various ML algorithms, it delivers results closely aligned with clinical practice, offering substantial clinical value for the guided use of VPA.

Exploring autism spectrum disorder and co-occurring trait associations to elucidate multivariate genetic mechanisms and insights

BackgroundAutism spectrum disorder (ASD) is a partially heritable neurodevelopmental trait, and people with ASD may also have other co-occurring trait such as ADHD, anxiety disorders, depression, mental health issues, learning difficulty, physical health traits and communication challenges. The concomitant development of ASD and other neurological traits is assumed to result from a complex interplay between genetics and the environment. However, only a limited number of studies have performed multivariate genome-wide association studies (GWAS) for ASD.MethodsWe conducted to-date the largest multivariate GWAS on ASD and 8 ASD co-occurring traits (ADHD, ADHD childhood, anxiety stress (ASDR), bipolar (BIP), disruptive behaviour (DBD), educational attainment (EA), major depression, and schizophrenia (SCZ)) using summary statistics from leading studies. Multivariate associations and central traits were further identified. Subsequently, colocalization and Mendelian randomization (MR) analysis were performed on the associations identified with the central traits containing ASD. To further validate our findings, pathway and quantified trait loci (QTL) resources as well as independent datasets consisting of 112 (45 probands) whole genome sequence data from the GEMMA project were utilized.ResultsMultivariate GWAS resulted in 637 significant associations (p < 5e-8), among which 322 are reported for the first time for any trait. 37 SNPs were identified to contain ASD and one or more traits in their central trait set, including variants mapped to known SFARI ASD genes MAPT, CADPS and NEGR1 as well as novel ASD genes KANSL1, NSF and NTM, associated with immune response, synaptic transmission, and neurite growth respectively. Mendelian randomization analyses found that genetic liability for ADHD childhood, ASRD and DBT has causal effects on the risk of ASD while genetic liability for ASD has causal effects on the risk of ADHD, ADHD childhood, BIP, WA, MDD and SCZ. Frequency differences of SNPs found in NTM and CADPS genes, respectively associated with neurite growth and neural/endocrine calcium regulation, were found between GEMMA ASD probands and controls. Pathway, QTL and cell type enrichment implicated microbiome, enteric inflammation, and central nervous system enrichments.ConclusionsOur study, combining multivariate GWAS with systematic decomposition, identified novel genetic associations related to ASD and ASD co-occurring driver traits. Statistical tests were applied to discern evidence for shared and interpretable liability between ASD and co-occurring traits. These findings expand upon the current understanding of the complex genetics regulating ASD and reveal insights of neuronal brain disruptions potentially driving development and manifestation.

Development of Vividness, Blackness, Depth, and Whiteness Scales Under High Range of Luminance Levels

ABSTRACTTwo‐dimensional (2D) color appearance scales, including attributes such as whiteness, blackness, vividness, and depth, are closely aligned with our daily experiences and are thus highly sought after. This article presents a psychophysical experiment utilizing the magnitude estimation method under four levels of luminance (10, 100, 1000, and 10 000 cd/m2). Observers evaluated both traditional one‐dimensional scales (brightness, colorfulness, and hue) and 2D scales (whiteness, blackness, vividness, and depth). The collected data were utilized to evaluate existing color appearance models, including CIECAM16, CAM16‐UCS, V*ab, and D*ab. Subsequently, new scales derived from the CAM16‐UCS model were developed to more accurately reflect the experimental data. These scales were further validated against independent data sets. The findings indicated that the scales developed from this study outperformed existing models when applied to external data sets. Moreover, Berns's scales for vividness and depth demonstrated strong predictive accuracy. Finally, four simple and accurate 2D scales are proposed.

Application of spatial metrology models in cell molecular localization and functional prediction

Understanding a protein’s exact cellular location is often essential to understanding its function. Even with the advancements in computer approaches, protein localization prediction indeed faces major obstacles such as interpretability and handling numerous localization sites. In this research, a novel approach, Squirrel Search Optimized Dynamic Visual Geometry Group Network (SSO-DVGG), is proposed to improve protein sub-cellular localization predictions by utilizing spatial metrology models to tackle these problems. With its simplified architecture, SSO-DVGG can explain whether a protein is directed to particular cellular sites, as well as identify important sequence components like sorting motifs or localization signals. This model allows users to select acceptable error levels by providing a confidence estimate for each prediction and highlighting sequence properties that are responsible for localization. This makes the model interpretable. Furthermore, SSO-DVGG uses a probabilistic methodology and integrates a large amount of data from dual-targeted proteins, which enables it to predict multiple localization locations per protein accurately. SSO-DVGG outperforms the best predictors and shows superior capacity to predict multiple localizations when tested on several independent datasets. By providing a clear and accurate understanding of protein distribution and function, this method promotes the application of spatial metrology models in cell molecular localization and functional prediction.

Single-cell transcriptomic and neuropathologic analysis reveals dysregulation of the integrated stress response in progressive supranuclear palsy

Progressive supranuclear palsy (PSP) is a sporadic neurodegenerative tauopathy variably affecting brainstem and cortical structures, and characterized by tau inclusions in neurons and glia. The precise mechanism whereby these protein aggregates lead to cell death remains unclear. To investigate the contribution of these different cellular abnormalities to PSP pathogenesis, we performed single-nucleus RNA sequencing (snRNA-seq) and analyzed 50,708 high quality nuclei targeting the diencephalon, including the subthalamic nucleus and adjacent structures, from human post-mortem PSP brains with varying degrees of pathology compared to controls. Cell-type-specific differential expression and pathway analysis identified both common and discrete changes in numerous pathways previously implicated in PSP and other neurodegenerative disorders. This included EIF2 signaling, an adaptive pathway activated in response to diverse stressors, which was activated in multiple vulnerable cell types and validated in independent snRNA-seq and bulk RNA-seq datasets. Using immunohistochemistry, we found that activated eIF2α was positively correlated with tau pathology burden in vulnerable brain regions. Multiplex immunofluorescence localized activated eIF2α positivity to hyperphosphorylated tau (p-tau) positive neurons and ALDH1L1-positive astrocytes, supporting the increased transcriptomic EIF2 activation observed in these vulnerable cell types. In conclusion, these data provide insights into cell-type-specific pathological changes in PSP and support the hypothesis that failure of adaptive stress pathways play a mechanistic role in the pathogenesis and progression of PSP.

Cell Type Differentiation Using Network Clustering Algorithms.

Single cell RNA-seq (scRNA-seq) technologies provide unprecedented resolution representing transcriptomics at the level of single cell. One of the biggest challenges in scRNA-seq data analysis is the cell type annotation, which is usually inferred by cell separation approaches. In-silico algorithms that accurately identify individual cell types in ongoing single-cell sequencing studies are crucial for unlocking cellular heterogeneity and understanding the biological basis of diseases. In this study, we focus on robustly identifying cell types in single-cell RNA sequencing data; we conduct a comparative analysis using methods established in biology, like Seurat, Leiden, and WGCNA, as well as Infomap, statistical inference via Stochastic Block Models (SBM), and single-cell Graph Neural Networks (scGNN). We also analyze preprocessing pipelines to identify and optimize key components in the process. Leveraging two independent datasets, PBMC and ROSMAP, we employ clustering algorithms on cell-cell networks derived from gene expression data. Our findings reveal that while clusters detected by WGCNA exhibit limited correspondence with cell types, those identified by multiresolution Infomap and Leiden, and SBM show a closer alignment, with Infomap standing out as a particularly effective approach. Infomap notably offers valuable insights for the precise characterization of cellular landscapes related to neurodegenration and immunology in scRNA-seq.

Prognostic significance of LOXL2 enzyme activity in primary myelofibrosis.

Primary myelofibrosis (PMF) is characterized by bone marrow fibrosis, but the underlying molecular mechanisms remain incompletely understood. Here, we investigated the role of lysyl oxidase-like 2 (LOXL2), an enzyme involved in extracellular matrix remodeling, in PMF pathogenesis. Analysis of bone marrow cells from PMF patients revealed significantly elevated LOXL2 mRNA expression compared to healthy controls, which was further validated using 2 independent Gene Expression Omnibus datasets (GSE26049 and GSE12234). Serum LOXL2 protein levels were markedly increased in PMF patients (n = 38) compared to controls (n = 15) (P < .01), accompanied by significantly enhanced enzymatic activity (P < .05). Clinical analysis demonstrated that high LOXL2 enzyme activity correlated with adverse clinical features, including higher risk scores in both International Prognostic Scoring System (P = .029) and Dynamic International Prognostic Scoring System (P = .012) systems, and increased peripheral blood blast percentages (≥1%, P = .019). Notably, elevated LOXL2 enzyme activity emerged as an independent adverse prognostic factor in multivariate analysis. Gene set enrichment analysis revealed LOXL2's association with multiple biological processes, including extracellular matrix remodeling, inflammatory responses, and JAK-STAT signaling pathways. Using CellMiner drug prediction analysis, we identified several FDA-approved drugs, particularly kinase inhibitors, as potential LOXL2-targeting therapeutics. Our findings demonstrate LOXL2's critical role in PMF pathogenesis and suggest its potential as both a prognostic marker and therapeutic target in PMF treatment.

Segmentation and Proportion Extraction of Crop, Crop Residues, and Soil Using Digital Images and Deep Learning

Conservation tillage involves covering the soil surface with crop residues after harvest, typically through reduced or no-tillage practices. This approach increases the soil organic matter, improves the soil structure, prevents erosion, reduces water loss, promotes microbial activity, and enhances root development. Therefore, accurate information on crop residue coverage is critical for monitoring the implementation of conservation tillage practices. This study collected “crop–crop residues–soil” images from wheat-soybean rotation fields using mobile phones to create calibration, validation, and independent validation datasets. We developed a deep learning model named crop–crop residue–soil segmentation network (CCRSNet) to enhance the performance of cropland “crop–crop residues–soil” image segmentation and proportion extraction. The model enhances the segmentation accuracy and proportion extraction by extracting and integrating shallow and deep image features and attention modules to capture multi-scale contextual information. Our findings indicated that (1) lightweight models outperformed deeper networks for “crop–crop residues–soil” image segmentation. When CCRSNet employed a deep network backbone (ResNet50), its feature extraction capability was inferior to that of lighter models (VGG16). (2) CCRSNet models that integrated shallow and deep features with attention modules achieved a high segmentation and proportion extraction performance. Using VGG16 as the backbone, CCRSNet achieved an mIoU of 92.73% and a PA of 96.23% in the independent validation dataset, surpassing traditional SVM and RF models. The RMSE for the proportion extraction accuracy ranged from 1.05% to 3.56%. These results demonstrate the potential of CCRSNet for the accurate, rapid, and low-cost detection of crop residue coverage. However, the generalizability and robustness of deep learning models depend on the diversity of calibration datasets. Further experiments across different regions and crops are required to validate this method’s accuracy and applicability for “crop–crop residues–soil” image segmentation and proportion extraction.

DeepAIPs-Pred: Predicting Anti-Inflammatory Peptides Using Local Evolutionary Transformation Images and Structural Embedding-Based Optimal Descriptors with Self-Normalized BiTCNs.

Inflammation is a biological response to harmful stimuli, playing a crucial role in facilitating tissue repair by eradicating pathogenic microorganisms. However, when inflammation becomes chronic, it leads to numerous serious disorders, particularly in autoimmune diseases. Anti-inflammatory peptides (AIPs) have emerged as promising therapeutic agents due to their high specificity, potency, and low toxicity. However, identifying AIPs using traditional in vivo methods is time-consuming and expensive. Recent advancements in computational-based intelligent models for peptides have offered a cost-effective alternative for identifying various inflammatory diseases, owing to their selectivity toward targeted cells with low side effects. In this paper, we propose a novel computational model, namely, DeepAIPs-Pred, for the accurate prediction of AIP sequences. The training samples are represented using LBP-PSSM- and LBP-SMR-based evolutionary image transformation methods. Additionally, to capture contextual semantic features, we employed attention-based ProtBERT-BFD embedding and QLC for structural features. Furthermore, differential evolution (DE)-based weighted feature integration is utilized to produce a multiview feature vector. The SMOTE-Tomek Links are introduced to address the class imbalance problem, and a two-layer feature selection technique is proposed to reduce and select the optimal features. Finally, the novel self-normalized bidirectional temporal convolutional networks (SnBiTCN) are trained using optimal features, achieving a significant predictive accuracy of 94.92% and an AUC of 0.97. The generalization of our proposed model is validated using two independent datasets, demonstrating higher performance with the improvement of ∼2 and ∼10% of accuracies than the existing state-of-the-art model using Ind-I and Ind-II, respectively. The efficacy and reliability of DeepAIPs-Pred highlight its potential as a valuable and promising tool for drug development and research academia.

Disentangling the consequences of type 2 diabetes on targeted metabolite profiles using causal inference and interaction QTL analyses.

Circulating metabolite levels have been associated with type 2 diabetes (T2D), but the extent to which T2D affects metabolite levels and their genetic regulation remains to be elucidated. In this study, we investigate the interplay between genetics, metabolomics, and T2D risk in the UK Biobank dataset using the Nightingale panel composed of 249 metabolites, 92% of which correspond to lipids (HDL, IDL, LDL, VLDL) and lipoproteins. By integrating these data with large-scale T2D GWAS from the DIAMANTE meta-analysis through Mendelian randomization analyses, we find 79 metabolites with a causal association to T2D, all spanning lipid-related classes except for Glucose and Tyrosine. Twice as many metabolites are causally affected by T2D liability, spanning almost all tested classes, including branched-chain amino acids. Secondly, using an interaction quantitative trait locus (QTL) analysis, we describe four metabolites consistently replicated in an independent dataset from the Estonian Biobank, for which genetic loci in two different genomic regions show attenuated regulation in T2D cases compared to controls. The significant variants from the interaction QTL analysis are significant QTLs for the corresponding metabolites in the general population but are not associated with T2D risk, pointing towards consequences of T2D on the genetic regulation of metabolite levels. Finally, through differential level analyses, we find 165 metabolites associated with microvascular, macrovascular, or both types of T2D complications, with only a few discriminating between complication classes. Of the 165 metabolites, 40 are not causally linked to T2D in either direction, suggesting biological mechanisms specific to the occurrence of complications. Overall, this work provides a map of the consequences of T2D on Nightingale targeted metabolite levels and on their genetic regulation, enabling a better understanding of the T2D trajectory leading to complications.

Deep learning-assisted colonoscopy images for prediction of mismatch repair deficiency in colorectal cancer.

Deficient mismatch repair or microsatellite instability is a major predictive biomarker for the efficacy of immune checkpoint inhibitors of colorectal cancer. However, routine testing has not been uniformly implemented due to cost and resource constraints. We developed and validated a deep learning-based classifiers to detect mismatch repair-deficient status from routine colonoscopy images. We obtained the colonoscopy images from the imaging database at Endoscopic Center of the Sixth Affiliated Hospital, Sun Yat-sen University. Colonoscopy images from a prospective trial (Neoadjuvant PD-1 blockade by toripalimab with or without celecoxib in mismatch repair-deficient or microsatellite instability-high locally advanced colorectal cancer) were used to test the model. A total of 5226 eligible images from 892 tumors from the consecutive patients were utilized to develop and validate the deep learning model. 2105 colorectal cancer images from 306 tumors were randomly selected to form model development dataset with a class-balanced approach. 3121 images of 488 proficient mismatch repair tumors and 98 deficient mismatch repair tumors were used to form the independent dataset. The model achieved an AUROC of 0.948 (95% CI 0.919-0.977) on the test dataset. On the independent validation dataset, the AUROC was 0.807 (0.760-0.854), and the NPV in was 94.2% (95% CI 0.918-0.967). On the prospective trial dataset, the model identified 29 tumors among the 33 deficient mismatch repair tumors (87.88%). The model achieved a high NPV in detecting deficient mismatch repair colorectal cancers. This model might serve as an automatic screening tool.

Plasma proteomic signatures of adiposity are associated with cardiovascular risk factors and type 2 diabetes risk in a multi-ethnic Asian population.

The biomarkers connecting obesity and cardiometabolic diseases are not fully understood. We aimed to (i) evaluate the associations between body mass index (BMI), waist circumference (WC), and ∼5,000 plasma proteins (SomaScan v4), (ii) identify protein signatures of BMI and WC, and (iii) evaluate the associations between the protein signatures and cardiometabolic health including metabolically unhealthy obesity and type 2 diabetes incidence in the Singapore Multi-Ethnic Cohort (MEC1). Among 410 BMI-associated and 385 WC-associated proteins, we identified protein signatures of BMI and WC and validated them in an independent dataset across two timepoints and externally in the Atherosclerosis Risk in Communities (ARIC) study. The BMI- and WC-protein signatures were highly correlated with total and visceral body fat, respectively. Furthermore, the protein signatures were significantly associated with cardiometabolic risk factors and metabolically unhealthy obesity. In prospective analyses, the protein signatures were strongly associated with type 2 diabetes risk in MEC1 (odds ratio per SD increment in WC-protein signature = 2.84, 95% CI 2.47 to 3.25) and ARIC (hazard ratio = 1.98, 95% CI 1.88 to 2.08). Our protein signatures have potential uses for the monitoring of metabolically unhealthy obesity.

ECMHA-PP: A Breast Cancer Prognosis Prediction Model Based on Energy-Constrained Multi-Head Self-Attention.

Breast cancer is a significant threat to women's health. Precise prognosis prediction for breast cancer can help doctors implement more rational treatment strategies. Artificial intelligence can assist doctors in decision-making and enhance predictionaccuracy. In this paper, a deep learning model ECMHA-PP (Energy Constrained Multi-Head Self-Attention based Prognosis Prediction) is proposed to predict the prognosis of breast cancer. ECMHA-PP utilizes patients' clinical data and extracts features through a cross-position mix and a channel mix multi-layer perceptron. Then, it incorporates an energy-constrained multi-head self-attention layer to improve feature extraction capability. The source code of ECMHA-PP has been hosted on GitHub and is available at https://github.com/xiaoliu166370/ECMHA-PP. To evaluate our proposed method, prognostic prediction experiments were performed on the METABRIC dataset, yielding outstanding results with an average accuracy of 93.0% and an average area under the curve of 0.974. To further validate the model's performance, we conducted tests on another independent dataset, BRCA, achieving an accuracy of 87.6%. In comparison with other widely used advanced methods, ECMHA-PP demonstrated higher comprehensive performance, making it a reliable prognostic prediction model for breast cancer. Given its robust feature extraction and predictioncapabilities.

Prediction of clinical prognosis and drug sensitivity in hepatocellular carcinoma through the combination of multiple cell death pathways.

Hepatocellular carcinoma (HCC) is the sixth most common malignant tumor, highlighting a significant need for reliable predictive models to assess clinical prognosis, disease progression, and drug sensitivity. Recent studies have highlighted the critical role of various programmed cell death pathways, including apoptosis, necroptosis, pyroptosis, ferroptosis, cuproptosis, entotic cell death, NETotic cell death, parthanatos, lysosome-dependent cell death, autophagy-dependent cell death, alkaliptosis, oxeiptosis, and disulfidptosis, in tumor development. Therefore, by investigating these pathways, we aimed to develop a predictive model for HCC prognosis and drug sensitivity. We analyzed transcriptome, single-cell transcriptome, genomic, and clinical information using data from the TCGA-LIHC, GSE14520, GSE45436, and GSE166635 datasets. Machine learning algorithms were used to establish a cell death index (CDI) with seven gene signatures, which was validated across three independent datasets, showing that high CDI correlates with poorer prognosis. Unsupervised clustering revealed three molecular subtypes of HCC with distinct biological processes. Furthermore, a nomogram integrating CDI and clinical information demonstrated good predictive performance. CDI was associated with immune checkpoint genes and tumor microenvironment components using single-cell transcriptome analysis. Drug sensitivity analysis indicated that patients with high CDI may be resistant to oxaliplatin and cisplatin but sensitive to axitinib and sorafenib. In summary, our model offers a precise prediction of clinical outcomes and drug sensitivity for patients with HCC, providing valuable insights for personalized treatment strategies.

DsRNAPredictor-II: An improved predictor of identifying dsRNA and its silencing efficiency for Tribolium castaneum based on sequence length distribution

RNA interference (RNAi) has been widely utilized to investigate gene functions and has significant potential for control of pest insects. However, recent studies have revealed that the target insect species, dsRNA molecule length, target genes, and other experimental factors can affect the efficiency of RNAi mediated control, restricting the further development and application of this technology. Therefore, the aim of this study was to establish a deep learning model using bioinformatics to help researchers identify dsRNA fragments with the highest RNAi efficiency. In this study, we optimized an existing model, namely, dsRNAPredictor, by designing sub-models based on different sequence lengths. Accordingly, the data were divided into two groups: 130–399 bp and 400–616 bp long sequences. Then, one-hot encoding was employed to extract sequence information. The convolutional neural network framework comprising three convolutional layers, three average pooling layers, a flattened layer, and three dense layers was employed as the classifier. By adjusting the parameters, we established two sub-models for different sequence distributions. Using multiple independent test datasets and conducting hypothesis testing, we demonstrated that our model exhibits superior performance and strong robustness to dsRNAPredictor, respectively. Therefore, our model may help design dsRNAs with pre-screening potential and facilitate further research and applications.

Assessing Internal Reproducibility Within a Parkinson’s Disease Cohort by Leveraging an Independent Larger Dataset

Assessing Internal Reproducibility Within a Parkinson’s Disease Cohort by Leveraging an Independent Larger Dataset

An Experimental Study of Byzantine-Robust Aggregation Schemes in Federated Learning

Byzantine-robust federated learning aims at mitigating Byzantine failures during the federated training process, where malicious participants may upload arbitrary local updates to the central server to degrade the performance of the global model. In recent years, several robust aggregation schemes have been proposed to defend against malicious updates from Byzantine clients and improve the robustness of federated learning. These solutions were claimed to be Byzantine-robust, under certain assumptions. Other than that, new attack strategies are emerging, striving to circumvent the defense schemes. However, there is a lack of systematic comparison and empirical study thereof. In this paper, we conduct an experimental study of Byzantine-robust aggregation schemes under different attacks using two popular algorithms in federated learning, FedSGD and FedAvg . We first survey existing Byzantine attack strategies and Byzantine-robust aggregation schemes that aim to defend against Byzantine attacks. We also propose a new scheme, ClippedClustering , to enhance the robustness of a clustering-based scheme by automatically clipping the updates. Then we provide an experimental evaluation of eight aggregation schemes in the scenario of five different Byzantine attacks. Our results show that these aggregation schemes sustain relatively high accuracy in some cases but are ineffective in others. In particular, our proposed ClippedClustering successfully defends against most attacks under independent and IID local datasets. However, when the local datasets are Non-IID, the performance of all the aggregation schemes significantly decreases. With Non-IID data, some of these aggregation schemes fail even in the complete absence of Byzantine clients. We conclude that the robustness of all the aggregation schemes is limited, highlighting the need for new defense strategies, in particular for Non-IID datasets.

An Interface for Linking Ancient Languages

Abstract This paper focuses on the linking potentials offered by the EpiLexO web-based front-end for creating and editing an ecosystem of digital resources for ancient languages, developed in the context of a project on the languages of fragmentary attestation of ancient Italy. The focus is particularly on mechanisms introduced for linking lexical information to other information bits either internally or externally, e.g., for creating attestations by linking lexical forms to their variants in relevant inscriptions, as well as for linking lexical data to external independent LOD datasets available on a remote endpoint. Finally, in the conclusions, we briefly introduce some future planned or desired enhancements as well as the final platform component, a parallel interface that constitutes the fruition application, which will be open to anyone on the web and will allow for browsing, searching, cross-querying and visualizing the created set of interlinked resources.