Biology subjects

Zhang, Q.

Publications and source records attributed to Zhang, Q..

At least 631 records · Page 35Linked to original sources

bioRxiv2020

Origins of 1/f-like tissue oxygenation fluctuations in the murine cortex

The concentration of oxygen in the brain spontaneously fluctuates, and the power distribution in these fluctuations has 1/f-like dynamics. Though these oscillations have been interpreted as being driven by neural activity, the origins of these 1/f-like oscillations is not well understood. Here, to gain insight of the origin of the 1/f-like oxygen fluctuations, we investigated the dynamics of tissue oxygenation and neural activity in awake behaving mice. We found that oxygen signal recorded from the cortex of mice had 1/f-like spectra. However, band-limited power in the local field potential, did not show corresponding 1/f-like fluctuations. When local neural activity was suppressed, the 1/f-like fluctuations in oxygen concentration persisted. Two-photon measurements of erythrocyte spacing fluctuations ( stalls) and mathematical modelling show that stochastic fluctuations in erythrocyte flow and stalling could underlie 1/f-like dynamics in oxygenation. These results show discrete nature of erythrocytes and their irregular flow, rather than neural activity, could drive 1/f-like fluctuations in tissue oxygenation.

neuroscience↗

bioRxiv2020

Codon-optimized FAM132b prevents diet-induced obesity by modulating adrenergic response and insulin action

FAM132b, also known as myonectin, has been identified as a myokine produced by exercise. It is a secreted protein precursor that belongs to the adipolin/erythroferrone family, and has hormone activity in circulation to regulate cellular iron homeostasis and lipid metabolism via unknown receptors. Here, adeno-associated viral vectors (AAV9) were engineered to induce overexpression of FAM132b with 2 codon mutations (A136T and P159A). Treatment of mice under high-fat diet feeding with FAM132b gene transfer resulted in marked reductions in body weight, fat depot, adipocytes size, glucose intolerance and insulin resistance. Moreover, FAM132b overproduction reduced glycemic response to epinephrine (EPI) in whole body and increased lipolytic response to EPI in adipose tissues. This adrenergic response of adipose tissue led to the result that gene transfer reduced glycogen utilization and increased fat consumption in skeletal muscle during exercise. FAM132b knockdown by shRNA significantly increased glycemic response to EPI in vivo and reduced adipocytes response to EPI and adipose tissue browning. Structural analysis suggested that FAM132b mutants delivered by AAV9 may form a weak bond with ADRB2, and potentially bind to insulin against insulin receptor by blocking the receptor binding sites on insulin B-chain. Our study underscores the potential of FAM132b gene therapy with codon optimization to treat obesity by modulating adrenergic response and interfering insulin action. SignificanceWe show here that AAV9-mediated expression of FAM132b with A136T and P159A is a safe and effective therapeutic strategy for improving glucose homeostasis. This is the first demonstration of a therapeutic effect on metabolic disorders in mice with FAM132b codon optimization. These therapeutic effects indicate that FAM132b gene transfer with selective codon mutants in vivo might be a valid therapy for diabetes that can be extended to other metabolic disorders.

molecular biology↗

bioRxiv2020

Genetic Diversity Analysis of Chinese Plums Based on Whole-genome Resequencing

Chinese plum (Prunus salicina L.), also known as Japanese plum, is gaining importance due to their extensive genetic diversity and nutritional attributes beneficial for human health. Single-nucleotide polymorphisms (SNPs) are the most abundant form of genomic polymorphisms and are widely used in population genetics research. Here, we construed high-density haplotype map by whole-genome resequencing of 67 Prunus accessions with a depth of ~20x to evaluate the genome-level diversity and population structure. The phylogenetic analysis, the principal component analysis, and the population structure profiling, indicated that the 67 plum accessions could be classified into four groups corresponding to their origin location, the southern cultivar group (SCG), the northern cultivar group (NCG), the foreign cultivar group (FG), and the mixed cultivar group (MG). Some cultivars from South China were clustered with the other three groups. The genetic diversity indices including the private allele number, the observed heterozygosity, the expected heterozygosity, and the nucleotide diversity of the SCG were higher than those of the NCG. The gene flow from the SCG to the FG was detected as well. We concluded that the origin center of Chinese plum was at the Yangtze River Basin in South China. This study provided genetic variation features and population structure of Chinese plum cultivars, laying a foundation for breeders to use diverse germplasm and allelic variants for improving Chinese plum varieties.

genetics↗

bioRxiv2020

StackPDB: predicting DNA-binding proteins based on XGB-RFE feature optimization and stacked ensemble classifier

DNA binding proteins (DBPs) not only play an important role in all aspects of genetic activities such as DNA replication, recombination, repair, and modification but also are used as key components of antibiotics, steroids, and anticancer drugs in the field of drug discovery. Identifying DBPs becomes one of the most challenging problems in the domain of proteomics research. Considering the high-priced and inefficient of the experimental method, constructing a detailed DBPs prediction model becomes an urgent problem for researchers. In this paper, we propose a stacked ensemble classifier based method for predicting DBPs called StackPDB. Firstly, pseudo amino acid composition (PseAAC), pseudo position-specific scoring matrix (PsePSSM), position-specific scoring matrix-transition probability composition (PSSM-TPC), evolutionary distance transformation (EDT), and residue probing transformation (RPT) are applied to extract protein sequence features. Secondly, extreme gradient boosting-recursive feature elimination (XGB-RFE) is employed to gain an excellent feature subset. Finally, the best features are applied to the stacked ensemble classifier composed of XGBoost, LightGBM, and SVM to construct StackPDB. After applying leave-one-out cross-validation (LOOCV), StackPDB obtains high ACC and MCC on PDB1075, 93.44% and 0.8687, respectively. Besides, the ACC of the independent test datasets PDB186 and PDB180 are 84.41% and 90.00%, respectively. The MCC of the independent test datasets PDB186 and PDB180 are 0.6882 and 0.7997, respectively. The results on the training dataset and the independent test dataset show that StackPDB has a great predictive ability to predict DBPs.

bioinformatics↗

bioRxiv2020

Statistics and classification of spatial rotations and projection directions considering molecular symmetry in 3D electron cryo-microscopy

Electron cryo-microscopy (cryoEM) three-dimensional (3D) reconstruction is based on estimations of orientations of projection images or 3D volumes. It is common that the macromolecules studied by cryoEM have molecular symmetry, which, unfortunately, has not been taken into consideration by any statistics for either spatial rotations or projection directions at this point. Meanwhile, there are growing needs to adopt advanced statistical methods, and further, modern machine learning techniques in cryoEM. Since those methodologies are built heavily upon statistical learning cornerstones, the absence of their domain-specific statistical justification limits their applications in cryoEM. In this research, based on the concept of non-unique-games (NUG), we propose two key statistical measurements, the mean and the variance, of both spatial rotations and projection directions when molecular symmetry is considered. Such methods are implemented in the open-source python package pySymStat.

biophysics↗

bioRxiv2020

Improved genetic prediction of complex traits from individual-level data or summary statistics

Most existing tools for constructing genetic prediction models begin with the assumption that all genetic variants contribute equally towards the phenotype. However, this represents a suboptimal model for how heritability is distributed across the genome. Therefore, we develop prediction tools that allow the user to specify the heritability model. We compare individual-level data prediction tools using 14 UK Biobank phenotypes; our new tool LDAK-Bolt-Predict outperforms the existing tools Lasso, BLUP, Bolt-LMM and BayesR for all 14 phenotypes. We compare summary statistic prediction tools using 225 UK Biobank phenotypes; our new tool LDAK-BayesR-SS outperforms the existing tools lassosum, sBLUP, LDpred and SBayesR for 223 of the 225 phenotypes. The increase in prediction accuracy from improving the heritability model tends to be substantial. For example, when using LDAK-Bolt-Predict, the proportion of phenotypic variance explained increased by on average 14% (range 5-29%), equivalent to increasing the sample size by a quarter.

genetics↗

bioRxiv2020

Recurrent human papillomavirus-related head and neck cancer undergoes metabolic re-programming and is driven by oxidative phosphorylation

Human papillomavirus (HPV) infection drives the development of some head and neck cancer squamous cell carcinomas (HNSC). This disease is rapidly increasing in incidence worldwide. Although these tumors are sensitive to treatment, ~10% of patients fail therapy. However, the mechanisms that underlie treatment failure remain unclear. Here, we show that the oxidative phosphorylation (OXPHOS) pathway is enriched in recurrent HPV-associated HNSC and may contribute to treatment failure. Nrf2-enriched HNSC samples from the Cancer Genome Atlas with enrichment in OXPHOS, fatty acid metabolism, Myc, Mtor, ROS, and glycolytic signaling networks exhibited worse survival. HPV-positive HNSC cells demonstrated sensitivity to the OXPHOS inhibitor, IACS-010759, in a Nrf2-dependent manner. Further, using murine xenograft models, we identified Nrf2 as a driver of tumor growth. Mechanistically, Nrf2 drives ROS and mitochondrial respiration, and Nrf2 is a critical regulator of redox homeostasis that can be crippled by disruption of OXPHOS. Nrf2 also mediated cisplatin sensitivity in endogenously overexpressing primary HPV-related HNSC cells. Cisplatin treatment demonstrated Nrf2-dependent synergy with OXPHOS inhibition. These results unveil a paradigm shifting translational target harnessing Nrf2-mediated metabolic reprogramming in HPV-related HNSC.

cancer biology↗

bioRxiv2020

Temporal Metabolic Characteristics and Transcriptomic Landscape of Islets and Liver Reveal Dynamic Pathophysiology and Interorgan Crosstalk in High-fat Diet-induced Diabetes

ObjectiveHyperinsulinemia and insulin resistance are co-existing characteristics of type 2 diabetes, whereas the molecular mechanism underlying this deleterious cycle remains elusive. The temporal transcriptomic landscape of core organs responsible for insulin secretion (islets) and insulin action (liver) could provide new insights. MethodsThe longitudinal profiling of glucose metabolism, insulin sensitivity, islet architecture and secretion were conducted in C57BL/6N mice fed a high-fat diet (HFD) or chow diet for 24 weeks. RNA-sequencing of islets and liver were performed once every 4 weeks. Weighted gene co-expression network analysis and Ingenuity Pathway Analysis were applied to construct networks and evaluate co-ordinated molecular interactions between islets and liver. ResultsMice exhibited progressively deteriorated glucose homeostasis with hyperinsulinemia but impaired first-phase insulin secretion after 4 weeks on HFD. Insulin, glucagon and somatostatin secretion in response to glucose with or without palmitate gradually deteriorated from dysregulation to failure. Systemic insulin resistance developed over 24 weeks with variable time course in tissue-specific insulin action. Our transcriptomic datasets outlined the impact of HFD on dynamics of islet and liver molecular network at different stages. Correlation analyses revealed that both organs jointly programmed {beta}-cell compensatory adaption via cell proliferation at early phase and irreversible islet dysfunction by inappropriate immune response at later stage. Alternations of T cell subpopulations validated the participation of adaptive immune response through priming and amplification phases in diabetic progression. ConclusionOur data provide a comprehensive landscape of crosstalk between islets and liver in diet-induced diabetes, elucidating the development of islet dysfunction and insulin resistance. HighlightsO_LIDiet-induced diabetes is featured by transition from islet dysfunction to failure C_LIO_LIInsulin resistance develops with variable time course in different tissues C_LIO_LIDynamics of islet and liver molecular network interplay at different stages C_LIO_LICell proliferation and improper immune reaction mediated interorgan crosstalk C_LIO_LIAdaptive immune response participated via priming and amplification phases C_LI

cell biology↗

bioRxiv2020

Investigation of white spot syndrome virus (WSSV) infection in wild crustaceans in the Bohai Sea

The ecological risks of white spot syndrome virus (WSSV), an important aquatic pathogen, has been causing increasing concern recently. A continuous survey on the prevalence of WSSV in the wild crustaceans of the Bohai Sea was conducted in present study. The result of loop-mediated isothermal amplification detection showed that WSSV positivity rates of sampling sites were determined to be 76.73%, 55.0% and 43.75% in 2016, 2017 and 2018, respectively. And the WSSV positivity rates of samples were 17.43%, 12.24% and 7.875% in 2016, 2017 and 2018, respectively. Meanwhile, the investigation revealed that 11 wild species from the sea were identified to be WSSV positive. The WSSV infection in wild crustacean species was confirmed by transmission electron microscopy analysis. The results of this study suggest that WSSV had been colonized in wild species offshore and the impact caused by WSSV to the wild marine ecosystem cannot be ignored.

microbiology↗

bioRxiv2020

Hematopoietic stem cells self-renew symmetrically or gradually proceed to differentiation

It is not known whether hematopoietic stem cells (HSCs) undergo symmetric or asymmetric cell divisions in the unperturbed bone marrow. Here, we integrate data from HSC fate mapping and cell-cycle-dependent labeling through mathematical inference and thus gain insight into how HSCs coordinate self-renewal with differentiation. We find that most HSC divisions in adult mice are symmetric self-renewing, replacing HSCs lost by direct differentiation and death, and slowly expanding the HSC population. This expansion maintains constant HSC output to multipotent progenitors (MPPs), despite declining HSC differentiation rate with age. We identify a linear hierarchy of differentiation states between tip HSCs and MPPs, where Tie2-driven HSC fate mapping fully covers the progression of the differentiating cells. A turning point from self-renewal to accelerated cell differentiation occurs between early-stage and late-stage MPPs, just before lineage differentiation becomes manifest in single-cell transcriptomes. This stem cell hierarchy precedes lineage differentiation and may limit mutation accumulation in the hematopoietic system.

systems biology↗

bioRxiv2020

Restoration of β-globin expression with optimally designed lentiviral vector for β-thalassemia treatment in Chinese patients

{beta}-thalassemia is one of the most prevalent genetic diseases worldwide. The current treatment for {beta}-thalassemia is allogeneic hematopoietic stem cell transplantation (HSCT), which is limited due to lack of matched donors. Gene therapy has been developed as an alternative therapeutic option for transfusion-ependent {beta}-thalassemia (TDT). However, successful gene therapy for {beta}-thalassemia patients in China has not been reported. Here we present the results of preclinical studies of an optimally designed LV named LentiHBBT87Q in hematopoietic stem cells (HSCs) derived from Chinese TDT patients. LentiHBBT87Q was selected from a series of LVs with optimized backbone and de novo cloning strategy. It contains an exogenous T87Q {beta}-globin gene (HBBT87Q) driven by a specific reconstituted locus control region (rLCR) and efficiently express HBB mRNA and HBB protein in erythroblasts derived from cord blood (CB) HSCs. To facilitate clinical transformation, we manufactured clinical grade LentiHBBT87Q (cLentiHBBT87Q) and optimized its transduction procedure. Importantly, transduction of cLentiHBBT87Qrestored expression of HBB monomer and adult hemoglobin (HbA) tetramer to relatively normal level in erythroblasts from bone marrow (BM) HSCs of Chinese TDT patients, that carry the most common mutation types and cover various genotypes, including {beta}0/{beta}0. Furthermore, viral integration sites (VIS) of cLentiHBBT87Q were similar to other LVs safely used in previous clinical trials and the associated risk of tumorigenesis was not observed in cLentiHBBT87Q transduced HSCs through comprehensive analysis. Taken together, we have engineered the cLentiHBBT87Q that can restore {beta}-globin expression in the HSCs-derived erythroblasts of Chinese TDT patients with minimal risk on tumorigenesis, providing a favorable starting point for future clinical application.

cell biology↗

bioRxiv2020

Visual Search P300 Source Analysis Based On ERP-fMRI Integration

Attention control can be achieved in two ways, stimulus-driven bottom-up attention and goal-driven top-down attention. Different visual search tasks involve different attention control. The pop-out task requires more bottom-up attention, whereas the search task involves more top-down attention. P300 which is the positive potential generated by the brain in the latency of 300-600 ms after the stimulus, reflects the processing of cognitive process and is an important component in visual attention. The P300 source is not consistent in the previous researches, our aim therefore, is to study the source location of P300 component based on visual search attention process. Here we use pop-out and search paradigm to get the ERP data of 13 subjects and the fMRI data of 25 subjects, and analyze the source location of P300 using the ERP-fMRI integration technology with high temporal resolution and high spatial resolution. The target differs from the distractor in color and orientation in the pop-out task, whereas the target and the distractor have different orientation and the same color in the search task. ERP results indicate that pop-out induces larger P300 concentrated in the parietal lobe, whereas search induced P300 is more distributed in the frontal lobe. Further ERP and fMRI integration analyses reveal that the left angular gyrus, right postcentral gyrus of parietal lobe and the left superior frontal gyrus (medial orbital) are the source of P300. Our study suggests the contribution of the frontal and parietal lobes to the P300 component.

neuroscience↗

bioRxiv2020

Targeting heparan sulfate proteoglycan-assisted endocytosis as a COVID-19 therapeutic option

The cell entry of SARS-CoV-2 has emerged as an attractive drug repurposing target for COVID-19. Here we combine genetics and chemical perturbation to demonstrate that ACE2-mediated entry of SARS-CoV and CoV-2 requires the cell surface heparan sulfate (HS) as an assisting cofactor: ablation of genes involved in HS biosynthesis or incubating cells with a HS mimetic both inhibit Spike-mediated viral entry. We show that heparin/HS binds to Spike directly, facilitates the attachment of viral particles to the cell surface to promote cell entry. We screened approved drugs and identified two classes of inhibitors that act via distinct mechanisms to target this entry pathway. Among the drugs characterized, Mitoxantrone is a potent HS inhibitor, while Sunitinib and BNTX disrupt the actin network to indirectly abrogate HS-assisted viral entry. We further show that drugs of the two classes can be combined to generate a synergized activity against SARS-CoV-2-induced cytopathic effect. Altogether, our study establishes HS as an attachment factor that assists SARS coronavirus cell entry, and reveals drugs capable of targeting this important step in the viral life cycle.

cell biology↗

bioRxiv2020

Binding Ability Prediction between Spike Protein and Human ACE2 Reveals the Adaptive Strategy of SARS-CoV-2 in Humans

SARS-CoV-2 (severe acute respiratory syndrome coronavirus 2) is a novel coronavirus causing an outbreak of COVID-19 globally in the past six months. A relatively higher divergence on the spike protein of SASR-CoV-2 enables it to transmit across species efficiently. We particularly believe that the adaptive mutations of the receptor-binding domain (RBD) of spike protein in SARS-CoV-2 might be essential to its high transmissibility among humans. Thus here we collected 2,142 high-quality genome sequences of SARS-CoV-2 from 160 regions in over 50 countries and reconstructed their phylogeny, and also analyzed the interaction between the polymorphisms of spike protein and human ACE2 (hACE2). Phylogenetic analysis of SARS-CoV-2 and coronavirus in other hosts show SARS-CoV-2 is highly possible originated from Bat-CoV (RaTG13) found in horseshoe bat and a recombination event may occur on the spike protein of Pangolin-CoV to imbue it the ability to infect humans. Moreover, compared to the S gene of SARS-CoV-2, it is more conserved in the direct-binding sites of RBD and we noticed that spike protein of SARS-CoV-2 may under a consensus evolution to adapt to human hosts better. 3,860 amino acid mutations in spike protein RBD (T333-C525) of SARS-CoV-2 were simulated and their stability and affinity binding to hACE2 (S19-D615) were calculated. Our analysis indicates SARS-CoV-2 could infect humans from different populations with no preference, and a higher divergence in the spike protein of SARS-CoV-2 at the early stage of this pandemic may be a good indicator that could show the pathway of SARS-CoV-2 transmitting from the natural reservoir to human beings.

microbiology↗

bioRxiv2020

The genome of North American beaver provides insights into the mechanisms of its longevity and cancer resistance

The North American beaver (Castor canadensis) is an exceptionally long-lived and cancer-resistant rodent species, and thus an excellent model organism for comparative genomic studies of longevity. Here, we utilize a significantly improved beaver genome assembly to assess evolutionary changes in gene coding sequences, copy number, and expression. We found that the beaver Aldh1a1, a stem cell marker gene encoding an enzyme required for detoxification of ethanol and aldehydes, is expanded (~10 copies vs. two in mouse and one in human). We also show that the beaver cells are more resistant to ethanol, and beaver liver extracts show higher ability to metabolize aldehydes than the mouse samples. Furthermore, Hpgd, a tumor suppressor gene, is uniquely duplicated in the beaver among rodents. Our evolutionary analysis identified beaver genes under positive selection which are associated with tumor suppression and longevity. Genes involved in lipid metabolism show positive selection signals, changes in copy number and altered gene expression in beavers. Several genes involved in DNA repair showed a higher expression in beavers which is consistent with the trend observed in other long-lived mammals. In summary, we identified several genes that likely contribute to beaver longevity and cancer resistance, including increased ability to detoxify aldehydes, enhanced tumor suppression and DNA repair, and altered lipid metabolism.

genomics↗

bioRxiv2020

Thiamine pyrophosphokinase deficiency induces Alzheimer's pathology

BackgroundThiamine diphosphate (TDP) reduction plays an important role in cerebral glucose hypometabolism, the neurodegenerative indicator, in Alzheimers disease (AD). The mechanism underlying TDP reduction remains elusive. Thus, it is critical to define the mechanism and its effect on neurodegeneration, the pathological basis of the disease occurrence and progression. MethodsThe mRNA levels of all known genes associated with thiamine metabolism, including thiamine pyrophosphokinase (TPK), Solute Carrier Family 19 Member 2 (SLC19A2), SLC19A3, and SLC25A19, in brain samples of patients with AD and other neurodegenerative disorders in multiple independent datasets were analyzed. TPK protein levels were further examined in the brain tissues of AD patients and control subjects. A mouse model with conditional knockout (cKO) of TPK gene in the excitatory neurons of adult brain was established. ResultsThe brain TPK mRNA level was markedly lower in AD patients, but not in other neurodegenerative disorders. The brain TPK protein level was also significantly decreased in AD patients. TPK gene knockout in the mice caused cerebral glucose hypometabolism, {beta}-amyloid deposition, Tau hyperphosphorylation, neuroinflammation, and neuronal loss and brain atrophy. Cross-species correlation analysis revealed the similar changes of gene profiling between the cKO mice and AD patients. ConclusionsThe deficiency of brain TPK, a key enzyme for TDP synthesis, is specific to AD. The cKO mice show AD-associated phenotypes and could serve as a new mouse model for AD studies. Our study provides a novel insight into the critical role of TPK in AD pathogenesis and its potential for the disease treatment.

neuroscience↗

bioRxiv2020

Convergent evolution between PALI1 and JARID2 for the allosteric activation of PRC2

The polycomb repressive complex 2 (PRC2) is a histone methyltransferase that maintains cell identities. JARID2 is the only accessory subunit of PRC2 that known to trigger an allosteric activation of methyltransferase. Yet, this mechanism cannot be generalised to all PRC2 variants as, in vertebrates, JARID2 is mutually exclusive with most of the accessory subunits of PRC2. Here we provide functional and structural evidence that the vertebrate-specific PRC2 accessory subunit PALI1 emerged through a convergent evolution to mimic JARID2 at the molecular level. Mechanistically, PRC2 methylates PALI1 K1241, which then binds to the PRC2-regulatory subunit EED to allosterically activate PRC2. PALI1 K1241 is methylated in mouse and human cell lines and is essential for PALI1-induced allosteric activation of PRC2. High-resolution crystal structures revealed that PALI1 mimics the regulatory interactions formed between JARID2 and EED. Independently, PALI1 also facilitates DNA and nucleosome binding by PRC2. In acute myelogenous leukemia cells, overexpression of PALI1 leads to cell differentiation, with the phenotype abrogated by a separation-of-function PALI1 mutation, defective in allosteric activation and active in DNA binding. Collectively, we show that PALI1 facilitates catalysis and substrate binding by PRC2 and provide evidence that subunit-induced allosteric activation is a general property of holo-PRC2 complexes.

biochemistry↗

bioRxiv2020

Morphologic Classification and Automatic Nugent Scoring of Bacterial Vaginosis by Deep Neural Networks

BackgroundBacterial vaginosis (BV) was the most common condition for womens health caused by the disruption of normal vaginal flora and an overgrowth of certain disease-causing bacteria, affecting 30-50% of women at some time in their lives. Gram stain followed by Nugent scoring (NS) based on bacterial morphotypes under the microscope was long considered golden standard for BV diagnosis. This conventional manual method was often considered labor intensive, time consuming, and variable results from person to person. MethodsWe developed four convolutional neural networks (CNN) models, and evaluated their ability to automatic identify vaginal bacteria and classify Nugent scores from microscope images. All the CNN models were first trained with 23280 microscopic images labeled with Nugent scores from top experts. A separate set of 5815 images were evaluated by the CNN models. The best CNN model was selected to generalize its application on an independent sets of 1082 images collecting from three teaching hospitals. Different hardwares were used to take images in hospitals. ResultsOur model could classify three Nugent Scores from images with high three classification accuracy of 89.3% (with 82.4% sensitivity and 96.6% specificity) on the 5815 test images, which was better diagnostic yield than the top-level technologists and obstetricians in China. The ability of generalization for our model was strong that it obtained 75.1%, which was 6.6% higher than the average of technologists. ConclusionThe CNN model over performed human healthcare practitioners on accuracy, efficiency and stability for BV diagnosis using microscopic image-based Nugent scores. The deep learning model may offer translational application in automating diagnosis of bacterial vaginosis with proper supporting hardware.

microbiology↗