PPML-Omics: A privacy-preserving federated machine learning method protects patients' privacy in omic data.

Modern machine learning models toward various tasks with omic data analysis give rise to threats of privacy leakage of patients involved in those datasets. Here, we proposed a secure and privacy-preserving machine learning method (PPML-Omics) by designing a decentralized differential private federated learning algorithm. We applied PPML-Omics to analyze data from three sequencing technologies and addressed the privacy concern in three major tasks of omic data under three representative deep learning models. We examined privacy breaches in depth through privacy attack experiments and demonstrated that PPML-Omics could protect patients' privacy. In each of these applications, PPML-Omics was able to outperform methods of comparison under the same level of privacy guarantee, demonstrating the versatility of the method in simultaneously balancing the privacy-preserving capability and utility in omic data analysis. Furthermore, we gave the theoretical proof of the privacy-preserving capability of PPML-Omics, suggesting the first mathematically guaranteed method with robust and generalizable empirical performance in protecting patients' privacy in omic data.

The association between ambient temperature and semen quality in a Northern Peninsular Province, China.

BACKGROUND: Extreme ambient temperature has been linked to decline in males' semen quality. Although the temperature-semen quality association has been examined in certain cities of South China, how the effect size of the extreme temperature may lag over critical windows in spermatogenesis and whether the strength of association may vary in North China have yet been adequately explored. OBJECTIVES: To quantify the association between air temperature and semen quality, and identify critical exposure windows in a Northern Peninsular Province, China. MATERIALS AND METHODS: Data on semen quality in 2014-2019 were collected from the Human Sperm Bank of Institute of Women, Children and Reproductive Health, Shandong University, China. Daily meteorological data (0.01°×0.01°) were assigned to each subject's residential address. The linear mixed-effect model combined with the distributed lag nonlinear model was used to estimate the effect of ambient temperature over critical spermatogenesis windows. RESULTS: The temperature-semen quality association was inverted V-shaped, with the maximum lag being 0-45 days before ejaculation and the threshold being 9.2°C. Progressively and total motile sperm number, and total sperm number declined more substantially than other semen quality parameters. Semen quality was more sensitive to cold exposure during the epididymal storage period than the sperm motility development period. By contrast, semen quality was insensitive to heat exposure during both critical spermatogenesis windows. Impairment of certain semen quality parameters was more obvious for males with higher educational attainment and those aged over 35 years. DISCUSSION AND CONCLUSION: Exposure to non-optimal temperature is associated with decreased semen quality in North China, with the epididymal storage and sperm motility development periods more sensitive to cold exposure than heat. Older males and those with higher educations may need particular awareness.

Personalized and privacy-preserving federated heterogeneous medical image analysis with PPPML-HMI.

Heterogeneous data is endemic due to the use of diverse models and settings of devices by hospitals in the field of medical imaging. However, there are few open-source frameworks for federated heterogeneous medical image analysis with personalization and privacy protection without the demand to modify the existing model structures or to share any private data. Here, we proposed PPPML-HMI, a novel open-source learning paradigm for personalized and privacy-preserving federated heterogeneous medical image analysis. To our best knowledge, personalization and privacy protection were discussed simultaneously for the first time under the federated scenario by integrating the PerFedAvg algorithm and designing the novel cyclic secure aggregation with the homomorphic encryption algorithm. To show the utility of PPPML-HMI, we applied it to a simulated classification task namely the classification of healthy people and patients from the RAD-ChestCT Dataset, and one real-world segmentation task namely the segmentation of lung infections from COVID-19 CT scans. Meanwhile, we applied the improved deep leakage from gradients to simulate adversarial attacks and showed the strong privacy-preserving capability of PPPML-HMI. By applying PPPML-HMI to both tasks with different neural networks, a varied number of users, and sample sizes, we demonstrated the strong generalizability of PPPML-HMI in privacy-preserving federated learning on heterogeneous medical images.

Effect of SARS-CoV-2 infection on semen parameters in sperm bank volunteers with normal sperm concentration.

In this study, we aimed to assess the effect of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection on semen parameters. The study comprised 110 sperm volunteers who self-reported SARS-CoV-2 infection from the Human Sperm Bank of the Center for Reproductive Medicine, Shandong University (Jinan, China). The volunteers had normal sperm concentration before infection. Each volunteer provided semen samples before and after infection. We selected 90 days after infection as the cutoff point. Semen parameters within 90 days after infection of 109 volunteers (group A) were compared with semen parameters before infection. Moreover, semen parameters on or after 90 days after infection of 36 volunteers (group B) were compared with semen parameters before infection. Furthermore, based on whether the volunteers had completed the three-dose SARS-CoV-2 vaccination booster, volunteers in group A and B were further divided into two subgroups separately. Semen parameters were compared before and after infection in each subgroup. Our results showed that in this cohort population, the semen quality in volunteers with normal sperm concentrations before infection decreased after SARS-CoV-2 infection within 90 days, while the semen quality returned to preinfection levels after 90 days. The completion of a three-dose SARS-CoV-2 vaccination booster may exert a protective effect on semen quality after infection.

Predicting the antigenic evolution of SARS-COV-2 with deep learning.

The relentless evolution of SARS-CoV-2 poses a significant threat to public health, as it adapts to immune pressure from vaccines and natural infections. Gaining insights into potential antigenic changes is critical but challenging due to the vast sequence space. Here, we introduce the Machine Learning-guided Antigenic Evolution Prediction (MLAEP), which combines structure modeling, multi-task learning, and genetic algorithms to predict the viral fitness landscape and explore antigenic evolution via in silico directed evolution. By analyzing existing SARS-CoV-2 variants, MLAEP accurately infers variant order along antigenic evolutionary trajectories, correlating with corresponding sampling time. Our approach identified novel mutations in immunocompromised COVID-19 patients and emerging variants like XBB1.5. Additionally, MLAEP predictions were validated through in vitro neutralizing antibody binding assays, demonstrating that the predicted variants exhibited enhanced immune evasion. By profiling existing variants and predicting potential antigenic changes, MLAEP aids in vaccine development and enhances preparedness against future SARS-CoV-2 variants.

Applications of deep learning in understanding gene regulation.

Gene regulation is a central topic in cell biology. Advances in omics technologies and the accumulation of omics data have provided better opportunities for gene regulation studies than ever before. For this reason deep learning, as a data-driven predictive modeling approach, has been successfully applied to this field during the past decade. In this article, we aim to give a brief yet comprehensive overview of representative deep-learning methods for gene regulation. Specifically, we discuss and compare the design principles and datasets used by each method, creating a reference for researchers who wish to replicate or improve existing methods. We also discuss the common problems of existing approaches and prospectively introduce the emerging deep-learning paradigms that will potentially alleviate them. We hope that this article will provide a rich and up-to-date resource and shed light on future research directions in this area.

Annotating TSSs in Multiple Cell Types Based on DNA Sequence and RNA-seq Data via DeeReCT-TSS.

The accurate annotation of transcription start sites (TSSs) and their usage are critical for the mechanistic understanding of gene regulation in different biological contexts. To fulfill this, specific high-throughput experimental technologies have been developed to capture TSSs in a genome-wide manner, and various computational tools have also been developed for in silico prediction of TSSs solely based on genomic sequences. Most of these computational tools cast the problem as a binary classification task on a balanced dataset, thus resulting in drastic false positive predictions when applied on the genome scale. Here, we present DeeReCT-TSS, a deep learning-based method that is capable of identifying TSSs across the whole genome based on both DNA sequence and conventional RNA sequencing data. We show that by effectively incorporating these two sources of information, DeeReCT-TSS significantly outperforms other solely sequence-based methods on the precise annotation of TSSs used in different cell types. Furthermore, we develop a meta-learning-based extension for simultaneous TSS annotations on 10 cell types, which enables the identification of cell type-specific TSSs. Finally, we demonstrate the high precision of DeeReCT-TSS on two independent datasets by correlating our predicted TSSs with experimentally defined TSS chromatin states. The source code for DeeReCT-TSS is available at https://github.com/JoshuaChou2018/DeeReCT-TSS_release and https://ngdc.cncb.ac.cn/biocode/tools/BT007316.

Self-supervised contrastive learning for integrative single cell RNA-seq data analysis.

We present a novel self-supervised Contrastive LEArning framework for single-cell ribonucleic acid (RNA)-sequencing (CLEAR) data representation and the downstream analysis. Compared with current methods, CLEAR overcomes the heterogeneity of the experimental data with a specifically designed representation learning task and thus can handle batch effects and dropout events simultaneously. It achieves superior performance on a broad range of fundamental tasks, including clustering, visualization, dropout correction, batch effect removal, and pseudo-time inference. The proposed method successfully identifies and illustrates inflammatory-related mechanisms in a COVID-19 disease study with 43 695 single cells from peripheral blood mononuclear cells.

Construction and validation of a prognostic model of RNA binding proteins in clear cell renal carcinoma.

BACKGROUND: The dysfunction of RNA binding proteins (RBPs) is associated with various inflammation and cancer. The occurrence and progression of tumors are closely related to the abnormal expression of RBPs. There are few studies on RBPs in clear cell renal carcinoma (ccRCC), which allows us to explore the role of RBPs in ccRCC. METHODS: We obtained the gene expression data and clinical data of ccRCC from the Cancer Genome Atlas (TCGA) database and extracted all the information of RBPs. We performed differential expression analysis of RBPs. Risk model were constructed based on the differentially expressed RBPs (DERBPs). The expression levels of model markers were examined by reverse transcription-quantitative PCR (RT-qPCR) and analyzed for model-clinical relevance. Finally, we mapped the model's nomograms to predict the 1, 3 and 5-year survival rates for ccRCC patients. RESULTS: The results showed that the five-year survival rate for the high-risk group was 40.2% (95% CI = 0.313 ~ 0.518), while the five-year survival rate for the low-risk group was 84.3% (95% CI = 0.767 ~ 0.926). The ROC curves (AUC = 0.748) also showed that our model had stable predictive power. Further RT-qPCR results were in accordance with our analysis (p < 0.05). The results of the independent prognostic analysis showed that the model could be an independent prognostic factor for ccRCC. The results of the correlation analysis also demonstrated the good predictive ability of the model. CONCLUSION: In summary, the 4-RBPs (EZH2, RPL22L1, RNASE2, U2AF1L4) risk model could be used as a prognostic indicator of ccRCC. Our study provides a possibility for predicting the survival of ccRCC.

Activating Lattice Oxygen in Layered Lithium Oxides through Cation Vacancies for Enhanced Urea Electrolysis.

Despite the fact that high-valent nickel-based oxides exhibit promising catalytic activity for the urea oxidation reaction (UOR), the fundamental questions concerning the origin of the high performance and the structure-activity correlations remain to be elucidated. Here, we unveil the underlying enhanced mechanism of UOR by employing a series of prepared cation-vacancy controllable LiNiO2 (LNO) model catalysts. Impressively, the optimized layered LNO-2 exhibits an extremely low overpotential at 10 mA cm-2 along with excellent stability after the 160 h test. Operando characterisations combined with the theoretical analysis reveal the activated lattice oxygen in layered LiNiO2 with moderate cation vacancies triggers charge disproportion of the Ni site to form Ni4+ species, facilitating deprotonation in a lattice oxygen involved catalytic process.

Role of neutrophil gelatinase-associated lipocalin in renal cell carcinoma.

Human neutrophil gelatinase-associated lipocalin (NGAL) is a glycoprotein present in a wide variety of tissues and cell types. It exists as a monomer of 25 kDa, a homodimer of 45 kDa or a heterodimer of 135 kDa (disulfide bound to latent matrix metalloproteinase-9). NGAL is considered the biochemical gold standard for the early diagnosis of acute kidney injury and has attracted much attention as a diagnostic biomarker. NGAL has controversial (i.e. both beneficial and detrimental) effects on cellular processes associated with tumor development, such as cell proliferation, survival, migration, invasion and drug resistance. Therefore, the present review aimed at clarifying the role of NGAL in renal cell carcinoma (RCC). Relevant studies of NGAL and RCC were searched in PubMed and relevant information about the structure, expression, function and mechanism of NGAL in RCC were summarized. Finally, the following conclusions could be drawn from the literature: i) NGAL can be detected in cancer tissues, serum and urine of patients with RCC; ii) NGAL is not a suitable diagnostic marker for early screening of RCC; iii) NGAL expression may be used to predict the prognosis of patients with RCC; and iv) Further research on NGAL may be helpful to decrease sunitinib resistance and find new treatment strategies for RCC.

HMD-ARG: hierarchical multi-task deep learning for annotating antibiotic resistance genes.

BACKGROUND: The spread of antibiotic resistance has become one of the most urgent threats to global health, which is estimated to cause 700,000 deaths each year globally. Its surrogates, antibiotic resistance genes (ARGs), are highly transmittable between food, water, animal, and human to mitigate the efficacy of antibiotics. Accurately identifying ARGs is thus an indispensable step to understanding the ecology, and transmission of ARGs between environmental and human-associated reservoirs. Unfortunately, the previous computational methods for identifying ARGs are mostly based on sequence alignment, which cannot identify novel ARGs, and their applications are limited by currently incomplete knowledge about ARGs. RESULTS: Here, we propose an end-to-end Hierarchical Multi-task Deep learning framework for ARG annotation (HMD-ARG). Taking raw sequence encoding as input, HMD-ARG can identify, without querying against existing sequence databases, multiple ARG properties simultaneously, including if the input protein sequence is an ARG, and if so, what antibiotic family it is resistant to, what resistant mechanism the ARG takes, and if the ARG is an intrinsic one or acquired one. In addition, if the predicted antibiotic family is beta-lactamase, HMD-ARG further predicts the subclass of beta-lactamase that the ARG is resistant to. Comprehensive experiments, including cross-fold validation, third-party dataset validation in human gut microbiota, wet-experimental functional validation, and structural investigation of predicted conserved sites, demonstrate not only the superior performance of our method over the state-of-art methods, but also the effectiveness and robustness of the proposed method. CONCLUSIONS: We propose a hierarchical multi-task method, HMD-ARG, which is based on deep learning and can provide detailed annotations of ARGs from three important aspects: resistant antibiotic class, resistant mechanism, and gene mobility. We believe that HMD-ARG can serve as a powerful tool to identify antibiotic resistance genes and, therefore mitigate their global threat. Our method and the constructed database are available at http://www.cbrc.kaust.edu.sa/HMDARG/ . Video abstract (MP4 50984 kb).

Correlations between glycolysis with clinical traits and immune function in bladder urothelial carcinoma.

BACKGROUND: Glycolysis was a representative hallmark in the tumor microenvironment (TME), and we aimed to explore the correlations between glycolysis with immune activity and clinical traits in bladder urothelial carcinoma (BLCA). METHODS: Our study obtained glycolysis scores for each BLCA samples from TCGA by a single-sample gene set enrichment analysis (ssGSEA) algorithm, based on a glycolytic gene set. The relationship between glycolysis with prognosis, clinical characteristics, and immune function were investigated subsequently. RESULTS: We found that enhanced glycolysis was associated with poor prognosis and metastasis in BLCA. Moreover, glycolysis had a close correlation with immune function, and enhanced glycolysis increased immune activities. In other words, glycolysis had a positive correlation with immune activities. Immune checkpoints such as IDO1, CD274, were up-regulated in high-glycolysis group as well. CONCLUSION: We speculated that in BLCA, elevated glycolysis enhanced immune function, which caused tumor cells to overexpress immune checkpoints to evade immune surveillance. Inhibition of glycolysis might be a promising assistant for immunotherapy in bladder cancer.

Recessive, Deleterious Variants in SMG8 Expand the Role of Nonsense-Mediated Decay in Developmental Disorders in Humans.

We have previously described a heart-, eye-, and brain-malformation syndrome caused by homozygous loss-of-function variants in SMG9, which encodes a critical component of the nonsense-mediated decay (NMD) machinery. Here, we describe four consanguineous families with four different likely deleterious homozygous variants in SMG8, encoding a binding partner of SMG9. The observed phenotype greatly resembles that linked to SMG9 and comprises severe global developmental delay, microcephaly, facial dysmorphism, and variable congenital heart and eye malformations. RNA-seq analysis revealed a general increase in mRNA expression levels with significant overrepresentation of core NMD substrates. We also identified increased phosphorylation of UPF1, a key SMG1-dependent step in NMD, which most likely represents the loss of SMG8--mediated inhibition of SMG1 kinase activity. Our data show that SMG8 and SMG9 deficiency results in overlapping developmental disorders that most likely converge mechanistically on impaired NMD.

Establishment and Validation of a Prognostic Risk Model for Autophagy-Related Genes in Clear Cell Renal Cell Carcinoma.

BACKGROUND: Autophagy plays an essential role in tumorigenesis. At present, due to the unclear role of autophagy in renal clear cell carcinoma, we studied the potential value of autophagy-related genes (ARGs) in renal clear cell carcinoma (ccRCC). METHODS: We obtained all ccRCC data from The Cancer Genome Atlas (TCGA). We extracted the expression data of ARGs for difference analysis and carried out biological function analysis on the different results. The autophagy risk model was constructed. The 5-year survival rate was assessed using the model, and the predictive power of the model was evaluated from multiple perspectives. Cox regression analysis was use to assess whether the model could be an independent prognostic factor. Finally, the correlation between the model and clinical indicators is analyzed. RESULTS: The patients were divided into the high-risk group and low-risk group according to the median of autophagy risk score, and the results showed that the prognosis of the low-risk group was better than that of a high-risk group. The validation results of external data sets show that our model has good predictive value for ccRCC patients. The model can be an independent prognostic factor. Finally, the results show that our model has a stable predictive ability. CONCLUSION: The autophagy gene model we constructed can be used as an excellent prognostic indicator for ccRCC. Our study provides the possibility of individualized and precise treatment for ccRCC patients.

Partial S substitution activates NiMoO4 for efficient and stable electrocatalytic urea oxidation.

Achieving a profound understanding of the reaction kinetics of a catalyst by modulating its electronic structure is significant. Herein, we present a scalable approach to achieving a spatially partial substitution of S into NiMoO4. The increase in active components in a true Ni3+ oxidation state as a result of optimizing the coordination environment greatly improved urea oxidation activity.