The next-generation K-means algorithm

Automated identification of fleck lesions in Stargardt disease using deep learning enhances lesion detection sensitivity and enables morphometric analysis of flecks

PURPOSE

To classify fleck lesions and assess artificial intelligence (AI) in identifying flecks in Stargardt disease (STGD).

METHODS

A retrospective study of 170 eyes from 85 consecutive patients with confirmed STGD. Fundus autofluorescence images were extracted, and flecks were manually outlined. A deep learning model was trained, and a hold-out testing subset was used to compare with manually identified flecks and for graders to assess. Flecks were clustered using K-means clustering.

RESULTS

Of the 85 subjects, 45 were female, and the median age was 37 years (IQR 25-59). A subset of subjects (n=41) had clearly identifiable fleck lesions, and an AI was successfully trained to identify these lesions (average Dice score of 0.53, n=18). The AI segmentation had smaller (0.018 compared with 0.034 mm2, p<0.001) but more numerous flecks (75.5 per retina compared with 40.0, p<0.001), but the total size of flecks was not different. The AI model had higher sensitivity to detect flecks but resulted in more false positives. There were two clusters of flecks based on morphology: broadly, one cluster of small round flecks and another of large amorphous flecks. The per cent frequency of small round flecks negatively correlated with subject age (r=-0.31, p<0.005).

CONCLUSIONS

AI-based detection of flecks shows greater sensitivity than human graders but with a higher false-positive rate. With further optimisation to address current shortcomings, this approach could be used to prescreen subjects for clinical research. The feasibility and utility of quantifying fleck morphology in conjunction with AI-based segmentation as a biomarker of progression require further study.

Differential Transcriptomic Landscapes of SARS-CoV-2 Variants in Multiple Organs from Infected Rhesus Macaques

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) caused the persistent coronavirus disease 2019 (COVID-19) pandemic, which has resulted in millions of deaths worldwide and brought an enormous public health and global economic burden. The recurring global wave of infections has been exacerbated by growing variants of SARS-CoV-2. In this study, the virological characteristics of the original SARS-CoV-2 strain and its variants of concern (VOCs; including Alpha, Beta, and Delta) in vitro, as well as differential transcriptomic landscapes in multiple organs (lung, right ventricle, blood, cerebral cortex, and cerebellum) from the infected rhesus macaques, were elucidated. The original strain of SARS-CoV-2 caused a stronger innate immune response in host cells, and its VOCs markedly increased the levels of subgenomic RNAs, such as N, Orf9b, Orf6, and Orf7ab, which are known as the innate immune antagonists and the inhibitors of antiviral factors. Intriguingly, the original SARS-CoV-2 strain and Alpha variant induced larger alteration of RNA abundance in tissues of rhesus monkeys than Beta and Delta variants did. Moreover, a hyperinflammatory state and active immune response were shown in the right ventricles of rhesus monkeys by the up-regulation of inflammation- and immune-related RNAs. Furthermore, peripheral blood may mediate signaling transmission among tissues to coordinate the molecular changes in the infected individuals. Collectively, these data provide insights into the pathogenesis of COVID-19 at the early stage of infection by the original SARS-CoV-2 strain and its VOCs.

The man, the plant, and the insect: shooting host specificity determinants in Serratia marcescens pangenome

Introduction

Serratia marcescens is most commonly known as an opportunistic pathogen causing nosocomial infections. It, however, was shown to infect a wide range of hosts apart from vertebrates such as insects or plants as well, being either pathogenic or growth-promoting for the latter. Despite being extensively studied in terms of virulence mechanisms during human infections, there has been little evidence of which factors determine S. marcescens host specificity. On that account, we analyzed S. marcescens pangenome to reveal possible specificity factors.

Methods

We selected 73 high-quality genome assemblies of complete level and reconstructed the respective pangenome and reference phylogeny based on core genes alignment. To find an optimal pipeline, we tested current pangenomic tools and obtained several phylogenetic inferences. The pangenome was rich in its accessory component and was considered open according to the Heaps' law. We then applied the pangenome-wide associating method (pan-GWAS) and predicted positively associated gene clusters attributed to three host groups, namely, humans, insects, and plants.

Results

According to the results, significant factors relating to human infections included transcriptional regulators, lipoproteins, ABC transporters, and membrane proteins. Host preference toward insects, in its turn, was associated with diverse enzymes, such as hydrolases, isochorismatase, and N-acetyltransferase with the latter possibly exerting a neurotoxic effect. Finally, plant infection may be conducted through type VI secretion systems and modulation of plant cell wall synthesis. Interestingly, factors associated with plants also included putative growth-promoting proteins like enzymes performing xenobiotic degradation and releasing ammonium irons. We also identified overrepresented functional annotations within the sets of specificity factors and found that their functional characteristics fell into separate clusters, thus, implying that host adaptation is represented by diverse functional pathways. Finally, we found that mobile genetic elements bore specificity determinants. In particular, prophages were mainly associated with factors related to humans, while genetic islands-with insects and plants, respectively.

Discussion

In summary, functional enrichments coupled with pangenomic inferences allowed us to hypothesize that the respective host preference is carried out through distinct molecular mechanisms of virulence. To the best of our knowledge, the presented research is the first to identify specific genomic features of S. marcescens assemblies isolated from different hosts at the pangenomic level.

scm6A-seq reveals single-cell landscapes of the dynamic m6A during oocyte maturation and early embryonic development

N⁶-methyladenosine (m⁶A) has been demonstrated to regulate RNA metabolism and various biological processes, including gametogenesis and embryogenesis. However, the landscape and function of m⁶A at single cell resolution have not been extensively studied in mammalian oocytes or during pre-implantation. In this study, we developed a single-cell m⁶A sequencing (scm⁶A-seq) method to simultaneously profile the m⁶A methylome and transcriptome in single oocytes/blastomeres of cleavage-stage embryos. We found that m⁶A deficiency leads to aberrant RNA clearance and consequent low quality of Mettl3^Gdf9 conditional knockout (cKO) oocytes. We further revealed that m⁶A regulates the translation and stability of modified RNAs in metaphase II (MII) oocytes and during oocyte-to-embryo transition, respectively. Moreover, we observed m⁶A-dependent asymmetries in the epi-transcriptome between the blastomeres of two-cell embryo. scm⁶A-seq thus allows in-depth investigation into m⁶A characteristics and functions, and the findings provide invaluable single-cell resolution resources for delineating the underlying mechanism for gametogenesis and early embryonic development.

Differential transcriptomic landscapes of multiple organs from SARS-CoV-2 early infected rhesus macaques

SARS-CoV-2 infection causes complicated clinical manifestations with variable multi-organ injuries, however, the underlying mechanism, in particular immune responses in different organs, remains elusive. In this study, comprehensive transcriptomic alterations of 14 tissues from rhesus macaque infected with SARS-CoV-2 were analyzed. Compared to normal controls, SARS-CoV-2 infection resulted in dysregulation of genes involving diverse functions in various examined tissues/organs, with drastic transcriptomic changes in cerebral cortex and right ventricle. Intriguingly, cerebral cortex exhibited a hyperinflammatory state evidenced by significant upregulation of inflammation response-related genes. Meanwhile, expressions of coagulation, angiogenesis and fibrosis factors were also up-regulated in cerebral cortex. Based on our findings, neuropilin 1 (NRP1), a receptor of SARS-CoV-2, was significantly elevated in cerebral cortex post infection, accompanied by active immune response releasing inflammatory factors and signal transmission among tissues, which enhanced infection of the central nervous system (CNS) in a positive feedback way, leading to viral encephalitis. Overall, our study depicts a multi-tissue/organ transcriptomic landscapes of rhesus macaque with early infection of SARS-CoV-2, and provides important insights into the mechanistic basis for COVID-19-associated clinical complications.

Unsupervised and semi-supervised learning: the next frontier in machine learning for plant systems biology

Advances in high-throughput omics technologies are leading plant biology research into the era of big data. Machine learning (ML) performs an important role in plant systems biology because of its excellent performance and wide application in the analysis of big data. However, to achieve ideal performance, supervised ML algorithms require large numbers of labeled samples as training data. In some cases, it is impossible or prohibitively expensive to obtain enough labeled training data; here, the paradigms of unsupervised learning (UL) and semi-supervised learning (SSL) play an indispensable role. In this review, we first introduce the basic concepts of ML techniques, as well as some representative UL and SSL algorithms, including clustering, dimensionality reduction, self-supervised learning (self-SL), positive-unlabeled (PU) learning and transfer learning. We then review recent advances and applications of UL and SSL paradigms in both plant systems biology and plant phenotyping research. Finally, we discuss the limitations and highlight the significance and challenges of UL and SSL strategies in plant systems biology.

Toxic effects of the combined cadmium and Cry1Ab protein exposure on the protective and transcriptomic responses of Pirata subpiraticus

Cadmium (Cd) pollution poses a serious threat to agricultural production and paddy field fauna. Crystalline proteins (e.g., Cry1Ab and Cry1Ac) are secreted by Bacillus thuringiensis, which can manage pests via a complicated toxic mechanism and have been widely used for pest control due to the commercialization of transgenic crops (e.g., cotton and rice) that expresses Bt insecticidal proteins. Nonetheless, studies on the effects of combined stress of Cd and Cry1Ab protein on field indicator species are limited. In the present study, we showed that spiders, Pirata subpiraticus, fed with Cd-containing flies+Cry1Ab had dramatically higher Cd accumulation than that in the spiders fed with Cd-containing flies (p < 0.05). In addition, the enrichment of Cd led to the activation of the protective mechanism by elevating the concentrations of glutathione peroxidase, glutathione S-transferase, and metallothionein in the spiders (p < 0.05). An in-depth transcriptome analysis revealed that the activities of ion metal binding proteins, transporters, and channels might play essential roles in the Cd accumulation process. More importantly, the higher Cd concentration in the combined Cd+Cry1Ab exposure prolonged developmental duration of P. subpiraticus, due to the down-regulated cuticle proteins (CPs) encoding genes involved in the molting process, which was regulated by a series of putative transcriptional factors such as ZBTB and zf-C2H2. Collectively, this integrated analysis illustrates that the combined Cd+Cry1Ab exposure increases the adverse effects of Cd stress on the growth, antioxidase, and CPs encoding genes of P. subpiraticus, thus providing a research basis and prospect for the rationality of transgenic Cry1Ab crops in the cultivation of heavy metal contaminated soil.

Monitoring of SARS-CoV-2 seroprevalence among primary healthcare patients in the Barcelona Metropolitan Area: the SeroCAP sentinel network protocol

INTRODUCTION

SARS-CoV-2 seroprevalence studies are currently being recommended and implemented in many countries. Forming part of the COVID-19 monitoring and evaluation plan of the Catalan Government Health Department, our network aims to initiate a primary healthcare sentinel monitoring system as a surrogate of SARS-CoV-2 exposure in the Barcelona Metropolitan Area.

METHODS AND ANALYSIS

The seroCAP is a serial cross-sectional study, which will be performed in the Barcelona Metropolitan Area to estimate antibodies against SARS-CoV-2. From February 2021 to March 2022, the detection of serum IgG antibodies against SARS-CoV-2 trimeric spike protein will be performed on a monthly basis in blood samples collected for diverse clinical purposes in three reference hospitals from the three Barcelona healthcare areas (BCN areas). The samples (n=2588/month) will be from patients attended by 30 primary healthcare teams at 30 basic healthcare areas (BHA). A lab software algorithm will systematically select the samples by age and sex. Seroprevalence will be estimated and monitored by age, sex, BCN area and BHA. Descriptive and cluster analysis of the characteristics and distribution of SARS-CoV-2 infections will be performed. Sociodemographic, socioeconomic and morbidity-associated factors will be determined using logistic regression. We will explore the association between seroprevalence, SARS-CoV-2 confirmed cases and the implemented measures using interrupted time series analysis.

ETHICS AND DISSEMINATION

Ethical approval was obtained from the University Institute Foundation for Primary Health Care Research Jordi Gol i Gurina ethics committee. An informed consent is not required regarding the approval of the secondary use of biological samples within the framework of the COVID-19 pandemic. A report will be generated quarterly. The final analysis, conclusions and recommendations will be shared with the stakeholders and communicated to the general public. Manuscripts resulting from the network will be submitted for publication in peer-reviewed journals.

The Distribution of Several Genomic Virulence Determinants Does Not Corroborate the Established Serotyping Classification of Bacillus thuringiensis

Bacillus thuringiensis, commonly referred to as Bt, is an object of the lasting interest of microbiologists due to its highly effective insecticidal properties, which make Bt a prominent source of biologicals. To categorize the exuberance of Bt strains discovered, serotyping assays are utilized in which flagellin serves as a primary seroreactive molecule. Despite its convenience, this approach is not indicative of Bt strains' phenotypes, neither it reflects actual phylogenetic relationships within the species. In this respect, comparative genomic and proteomic techniques appear more informative, but their use in Bt strain classification remains limited. In the present work, we used a bottom-up proteomic approach based on fluorescent two-dimensional difference gel electrophoresis (2D-DIGE) coupled with liquid chromatography/tandem mass spectrometry(LC-MS/MS) protein identification to assess which stage of Bt culture, vegetative or spore, would be more informative for strain characterization. To this end, the proteomic differences for the israelensis-attributed strains were assessed to compare sporulating cultures of the virulent derivative to the avirulent one as well as to the vegetative stage virulent bacteria. Using the same approach, virulent spores of the israelensis strain were also compared to the spores of strains belonging to two other major Bt serovars, namely darmstadiensis and thuringiensis. The identified proteins were analyzed regarding the presence of the respective genes in the 104 Bt genome assemblies available at open access with serovar attributions specified. Of 21 proteins identified, 15 were found to be encoded in all the present assemblies at 67% identity threshold, including several virulence factors. Notable, individual phylogenies of these core genes conferred neither the serotyping nor the flagellin-based phylogeny but corroborated the reconstruction based on phylogenomics approaches in terms of tree topology similarity. In its turn, the distribution of accessory protein genes was not confined to the existing serovars. The obtained results indicate that neither gene presence nor the core gene sequence may serve as distinctive bases for the serovar attribution, undermining the notion that the serotyping system reflects strains' phenotypic or genetic similarity. We also provide a set of loci, which fit in with the phylogenomics data plausibly and thus may serve for draft phylogeny estimation of the novel strains.

A Novel Machine Learning Framework for Comparison of Viral COVID-19-Related Sina Weibo and Twitter Posts: Workflow Development and Content Analysis

Background

Social media plays a critical role in health communications, especially during global health emergencies such as the current COVID-19 pandemic. However, there is a lack of a universal analytical framework to extract, quantify, and compare content features in public discourse of emerging health issues on different social media platforms across a broad sociocultural spectrum.

Objective

We aimed to develop a novel and universal content feature extraction and analytical framework and contrast how content features differ with sociocultural background in discussions of the emerging COVID-19 global health crisis on major social media platforms.

Methods

We sampled the 1000 most shared viral Twitter and Sina Weibo posts regarding COVID-19, developed a comprehensive coding scheme to identify 77 potential features across six major categories (eg, clinical and epidemiological, countermeasures, politics and policy, responses), quantified feature values (0 or 1, indicating whether or not the content feature is mentioned in the post) in each viral post across social media platforms, and performed subsequent comparative analyses. Machine learning dimension reduction and clustering analysis were then applied to harness the power of social media data and provide more unbiased characterization of web-based health communications.

Results

There were substantially different distributions, prevalence, and associations of content features in public discourse about the COVID-19 pandemic on the two social media platforms. Weibo users were more likely to focus on the disease itself and health aspects, while Twitter users engaged more about policy, politics, and other societal issues.

Conclusions

We extracted a rich set of content features from social media data to accurately characterize public discourse related to COVID-19 in different sociocultural backgrounds. In addition, this universal framework can be adopted to analyze social media discussions of other emerging health issues beyond the COVID-19 pandemic.