Biology subjects

Deng, F.

Publications and source records attributed to Deng, F..

At least 37 records · Page 2Linked to original sources

bioRxiv2024

Building a literature knowledge base towards transparent biomedical AI

As artificial intelligence (AI) continues to advance and scale up in biomedical research, concerns about AIs trustworthiness and transparency have grown. There is a critical need to systematically bring accurate and relevant biomedical knowledge into AI applications for transparency and provenance. Knowledge graphs have emerged as a powerful tool that integrates heterogeneous knowledge by explicitly describing biomedical knowledge as entities and relationships between entities. However, PubMed, the largest and most comprehensive repository of biomedical knowledge, exists primarily as unstructured text and is under utilized for advanced machine learning tasks. To address the challenge, we developed LiteralGraph, a computational framework to extract biomedical terms and relationships from PubMed literature into a unified knowledge graph. Using this framework, we established the Genomic Literature Knowledge Base (GLKB), which consolidates 14,634,427 biomedical relationships between 3,276,336 biomedical terms from over 33 million PubMed abstracts and nine well-established biomedical repositories. The database is coupled with RESTful APIs and a user-friendly web interface that makes it accessible to researchers for various usages. We demonstrated the broad utility of GLKB towards transparent AI in three distinct application scenarios. In the LLM grounding scenario, we developed a Retrieval Augmented Generation (RAG) agent to reduce LLM hallucination in biomedical question answering. In the hypothesis generation scenario, we elucidated the potential functions of RFX6 in type 2 diabetes (T2D) using the vast evidence from PubMed articles. In the machine learning scenario, we utilized GLKB to provide semantic knowledge in predictive tasks and scientific fact-checking.

bioinformatics↗

bioRxiv2024

Efficient Gene Editing and Overexpression of Gametophyte Transformation in a Model Fern

The clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-related nuclease (Cas) system allows precise and easy editing of genes in many plant species. However, this system has not yet been applied to any fern species due to the complex characteristics of fern genomes, genetics and physiology. Here, we established, for the first time, a protocol for gametophyte-based screening single-guide RNAs (sgRNAs) with high efficiency for CRISPR/Cas-mediated gene editing in a model fern species, Ceratopteris richardii. We utilized the C. richardii Actin promoter to drive sgRNA expression and enhanced CaMV 35S promoter to drive the expression of Streptococcus pyogenes Cas9 in this CRISPR-mediated editing system, which was employed to successfully edit a few genes (e.g., nucleotidase/phosphatase 1, CrSAL1; Cryptochrome 4, CRY4) and CrPDS, encoding a phytoene desaturase protein that resulted in an albino phenotype in C. richardii. Knockout of CrSAL1 resulted in significantly reduced stomatal conductance (gs), leaf transpiration rate (E), stomatal/pore length, and abscisic acid (ABA)-induced reactive oxygen species (ROS) accumulation in guard cells. Moreover, CrSAL1 overexpressing plants showed significantly increased net photosynthetic rate (A), gs, E and intrinsic water use efficiency (iWUE) as well as most of the stomatal traits and ROS production in guard cells compared to those in the wild-type (WT) plants. Taken together, the optimized CRISPR/Cas9 system provides a useful tool for functional genomics in a model fern species, allowing the exploration of fern gene functions for evolutionary biology, herbal medicine discovery and agricultural applications.

plant biology↗

bioRxiv2024

Prominent involvement of acetylcholine in shaping stable olfactory representation across the Drosophila brain

Despite the vital role of neuromodulation in the neural system, the specific spatiotemporal dynamics of neuromodulators and their interactions with neuronal activities in vivo are still unclear, hampering our understanding of their information representation and functional contributions systemically. To address this problem, we employed two-photon synthetic aperture microscopy (2pSAM) to record long-term neuronal and neuromodulatory olfactory responses across the Drosophila brain at high speed. Our results revealed distinct response properties, global information propagation, functional connectivity, and odor identity representation among neuronal, cholinergic, and serotoninergic dynamics across multiple brain regions. We discovered the compensation between neuronal activity and cholinergic dynamics, both in the odor identity representation across the brain and the functional connectivity network structures of specific brain regions. Moreover, employing low-dimensional manifold and functional connectivity network analyses, we characterized the stable representation of cholinergic dynamics over a long term. Collectively, our unbiased and comprehensive investigation unveiled the prominent involvement of acetylcholine (ACh) in shaping olfactory representation across the brain, underscoring the inadequacy of solely considering neuronal activities when examining information representation of the brain.

neuroscience↗

bioRxiv2024

Establishing Primary and Stable Cell Lines from Frozen Wing Biopsies for Cellular, Physiological, and Genetic Studies in Bats

Bats stand out among mammalian species for their exceptional traits, including the capacity to navigate through flight and echolocation, conserve energy through torpor/hibernation, harbor a multitude of viruses, exhibit resistance to disease, survive harsh environmental conditions, and demonstrate exceptional longevity compared to other mammals of similar size. In vivo studies of bats can be challenging for several reasons such as ability to locate and capture them in their natural environments, limited accessibility, low sample size, environmental variation, long lifespans, slow reproductive rates, zoonotic disease risks, species protection, and ethical concerns. Thus, establishing alternative laboratory models is crucial for investigating the diverse physiological adaptations observed in bats. Obtaining quality cells from tissues is a critical first step for successful primary cell derivation. However, it is often impractical to collect fresh tissue and process the samples immediately for cell culture due to the resources required for isolating and expanding cells. As a result, frozen tissue is typically the starting resource for bat primary cell derivation. Yet, cells in frozen tissue are usually damaged and represent low integrity and viability. As a result, isolating primary cells from frozen tissues poses a significant challenge. Herein, we present a successfully developed protocol for isolating primary dermal fibroblasts from frozen bat wing biopsies. This protocol marks a significant milestone, as this the first protocol specially focused on fibroblasts isolation from bat frozen tissue. We also describe methods for primary cell characterization, genetic manipulation of primary cells through lentivirus transduction, and the development of stable cell lines. Basic Protocol 1: Bat wing biopsy collection and preservation Support Protocol 1: Blood collection from bat-venipuncture Basic Protocol 2: Isolation of primary fibroblasts from adult bat frozen wing biopsy Support Protocol 2: Maintenance of primary fibroblasts Support Protocol 3: Cell banking and thawing of primary fibroblasts Support Protocol 4: Growth curve and doubling time Support Protocol 5: Lentiviral transduction of bat primary fibroblasts Basic Protocol 3: Bat stable fibroblasts cell lines development Support Protocol 6: Bat fibroblasts validation by immunofluorescence staining Support Protocol 7: Chromosome counting

cell biology↗

bioRxiv2024

Hairpin-locker mediated CRISPR/Cas tandem system for ultrasensitive detection of DNA without pre-amplification

Achieving ultra-sensitive detection of DNA is of paramount importance in the field of molecular analytics. Conventional amplification technologies such as polymerase chain reaction (PCR) currently play a leading role in ultrasensitive DNA detection. However, amplicon contamination common in these techniques may lead to false positives. To date, CRISPR-associated nucleases (type V & VI) with their programmable cleavage have been utilised for sensitive detection of unamplified nucleic acids in complex real samples. Nevertheless, without additional amplification strategies, the pM range sensitivity of such CRISPR/Cas sensors is not sufficient for clinical applications. Here, we established a hairpin-locker (H-locker) mediated Cas12-Cas13 tandem biosensing system (Cas12-13 tandem-sensor) for ultrasensitive detection of DNA targets. Without the need for any additional amplification reaction or device, this system is capable of detecting DNA at a notable 1 aM level (<1 copy/uL) sensitivity. In addition, the system was able to distinguish cancer mutations in colorectal cancer (CRC) mice. This is a significant advance for CRISPR/Cas biosensing technology offering simple, highly sensitive, and user-friendly diagnostics for next-generation nucleic acid detection. O_TBL View this table: org.highwire.dtl.DTLVardef@7fc32eorg.highwire.dtl.DTLVardef@1e716ceorg.highwire.dtl.DTLVardef@8a43e0org.highwire.dtl.DTLVardef@16e286dorg.highwire.dtl.DTLVardef@1202eb0_HPS_FORMAT_FIGEXP M_TBL O_FLOATNOTable 1.C_FLOATNO O_TABLECAPTIONDNA and RNA oligos used in this study. C_TABLECAPTION C_TBL

molecular biology↗

bioRxiv2024

Modulating Cas13a trans-cleavage by double strand RNA: Application to the development of an autocatalytic sensor

Cas13a-based diagnostic systems have been widely utilized for the detection of RNA targets. However, without preamplification such systems have sensitivity in the picomolar range only. Here, we found that double strand RNA (dsRNA) over 20nt is able to effectively activate the trans-cleavage activity of Cas13a RNP, while the cleavage rates of dsRNA by activated Cas13a RNP are very low. In addition, specially designed small circular RNA constructs (Cir-mediators comprising a 20nt dsRNA trigger with a 5nt ssRNA linker) have limited ability to activate Cas13a RNP, but this activation is restored once the circular structures are cleaved and become linear. Based on this new method to control trans-cleavage activity of Cas13 RNP, we developed a Cas13a autocatalytic biosensing system assisted by Cir-mediators, which allows one target RNA to activate numerous Cas13a RNPs. With this approach we show ultrasensitive detection of 1aM of synthetic RNA targets without preamplification within 15min. The sensor was successfully applied to monitor miRNA-21 concentration in clinical plasma samples in colorectal cancer. This investigation yields novel insights into the properties of Cas13a RNPs, and the Cir-mediator-based autosensor introduces a novel method for detecting RNA targets with exceptional sensitivity. HighlightsO_LIdsRNA is able to trigger activation of Cas13a RNP. C_LIO_LIActivated Cas13a RNP do not cleave dsRNA. C_LIO_LICir-mediator induces low levels of Cas13a RNP activation. C_LIO_LICir-mediator based Cas13a auto-catalysis biosensor can detect 1aM RNA targets. C_LI

molecular biology↗

bioRxiv2024

Crypt and Villus Enterochromaffin Cells are Distinct Stress Sensors in the Gut

The crypt-villus structure of the small intestine serves as an essential protective barrier, with its integrity monitored by the guts sensory system. Enterochromaffin (EC) cells, which are rare sensory epithelial cells that release serotonin (5-HT), surveil the mucosal environment and signal both within and outside the gut. However, it remains unclear whether EC cells in intestinal crypts and villi respond to different stimuli and elicit distinct responses. In this study, we introduce a new reporter mouse model to observe the release and propagation of serotonin in live intestines. Using this system, we show that crypt EC cells exhibit two modes of serotonin release: transient receptor potential A1 (TRPA1)-dependent tonic serotonin release that controls basal ionic secretion, and irritant-evoked serotonin release that activates gut sensory neurons. Furthermore, we find that a thick protective mucus layer prevents TRPA1 receptors on crypt EC cells from responding to luminal irritants such as reactive electrophiles; if this mucus layer is compromised, then crypt EC cells become susceptible to activation by luminal irritants. On the other hand, villus EC cells detect oxidative stress through TRPM2 channels and co-release serotonin and ATP to activate nearby gut sensory fibers. Our work highlights the physiological importance of intestinal architecture and differential TRP channel expression in sensing noxious stimuli that elicit nausea and/or pain sensations in the gut.

physiology↗

bioRxiv2023

LncPNdeep: A long non-coding RNA classifier based on Large Language Model with peptide and nucleotide embedding

Long non-coding RNA plays an important role in various gene transcription and peptide interactions. Classifying lncRNAs from coding RNA is a crucial step in bioinformatics analysis which seriously affects the post-analysis for transcriptome annotation. Although several machine learning-based methods were developed to classify lncRNAs, these methods were mainly focused on nucleotide features without considering the information from the peptide sequence. To integrate both nucleotide and peptide information in lncRNA classification, one efficient deep learning is desired. In this study, we developed one concatenated deep neural network named LncPNdeep to combine this information. LncPNdeep incorporates both peptide and nucleotide embedding from masked language modeling (MLM), being able to discover complex associations between sequence information and lncRNA classification. LncPNdeep achieves state-of-the-art performance in the human transcript database compared with other existing methods (Accuracy=97.1%). It also exhibits superior generalization ability in cross-species comparison, maintaining consistent accuracy and F1 scores compared to other methods. The combination of nucleotide and peptide information makes LncPNdeep able to facilitate the identification of novel lncRNA and gain high accuracy for classification. Our code is available at https://github.com/yatoka233/LncPNdeep

bioinformatics↗

bioRxiv2023

Monitoring norepinephrine release in vivo using next-generation GRABNE sensors

Norepinephrine (NE) is an essential biogenic monoamine neurotransmitter, yet researches using prototype NE sensors were limited by their low sensitivities. Here, we developed next-generation versions of GPCR activation-based NE sensors (GRABNE2m and GRABNE2h) with a superior response, high sensitivity and selectivity to NE both in vitro and in vivo. Notably, these sensors can detect NE release triggered by either optogenetic or behavioral stimuli in freely moving mice, producing robust signals in the locus coeruleus and hypothalamus. With the development of a novel transgenic mouse line, we recorded both NE release and calcium dynamics with dual-color fiber photometry throughout the sleep-wake cycle; moreover, dual-color mesoscopic imaging revealed cell type-specific spatiotemporal dynamics of NE and calcium during sensory processing and locomotion. Thus, these new GRABNE sensors are valuable tools for monitoring the precise spatiotemporal release of NE in vivo, providing new insights into the physiological and pathophysiological roles of NE.

neuroscience↗

bioRxiv2023

Dual-color GRAB sensors for monitoring spatiotemporal serotonin release in vivo

The serotonergic system plays important roles in both physiological and pathological processes, and is a widely used therapeutic target for many psychiatric disorders. Although several genetically encoded GFP-based serotonin (5-HT) sensors were recently developed, their sensitivities and spectral profiles are relatively limited. To overcome these limitations, we optimized green fluorescent G-protein-coupled receptor (GPCR)-activation-based 5-HT (GRAB5-HT) sensors and developed a new red fluorescent GRAB5-HT sensor. These sensors have excellent cell surface trafficking, high specificity, sensitivity, and spatiotemporal resolution, making them suitable for monitoring 5-HT dynamics in vivo. Besides recording subcortical 5-HT release in freely moving mice, we observed both uniform and gradient 5-HT release in the mouse dorsal cortex with mesoscopic imaging. Finally, we performed dual-color imaging and observed seizure-induced waves of 5-HT release throughout the cortex following calcium and endocannabinoid waves. In summary, these 5-HT sensors can offer valuable insights regarding the serotonergic system in both physiological and pathological states.

neuroscience↗

bioRxiv2023

Exomap1 mouse: a transgenic model for in vivo studies of exosome biology

Exosomes are small extracellular vesicles (sEVs) of [~]30-150 nm in diameter that have the same topology as the cell, are enriched in selected exosome cargo proteins, and play important roles in health and disease. To address large unanswered questions regarding exosome biology in vivo, we created the exomap1 transgenic mouse model. In response to Cre recombinase, exomap1 mice express HsCD81mNG, a fusion protein between human CD81, the most highly enriched exosome protein yet described, and the bright green fluorescent protein mNeonGreen. As expected, cell type-specific expression of Cre induced the cell type-specific expression of HsCD81mNG in diverse cell types, correctly localized HsCD81mNG to the plasma membrane, and selectively loaded HsCD81mNG into secreted vesicles that have the size ([~]80 nm), topology (outside out), and content (presence of mouse exosome markers) of exosomes. Furthermore, mouse cells expressing HsCD81mNG released HsCD81mNG-marked exosomes into blood and other biofluids. Using high-resolution, single-exosome analysis by quantitative single molecule localization microscopy, we show here that that hepatocytes contribute [~]15% of the blood exosome population whereas neurons contribute <1% of blood exosomes. These estimates of cell type-specific contributions to blood EV population are consistent with the porosity of liver sinusoidal endothelial cells to particles of [~]50-300 nm in diameter, as well as with the impermeability of blood-brain and blood-neuron barriers to particles >5 nm in size. Taken together, these results establish the exomap1 mouse as a useful tool for in vivo studies of exosome biology, and for mapping cell type-specific contributions to biofluid exosome populations. In addition, our data confirm that CD81 is a highly-specific marker for exosomes and is not enriched in the larger microvesicle class of EVs.

cell biology↗

bioRxiv2023

Molecular Evolution and Interaction of 14-3-3 Proteins with H+-ATPases in Plant Abiotic Stresses

Environmental stresses severely affect plant growth and crop productivity. Regulated by 14-3-3 proteins (14-3-3s), H+-ATPases (AHA) are important proton pumps that can induce diverse secondary transport via channels and co-transporters for the abiotic stress response of plants. Many studies demonstrated the roles of 14-3-3s and AHAs in coordinating the processes of plant growth, phytohormone signaling, and stress responses. However, the molecular evolution of 14-3-3s and AHAs has not been summarized in parallel with insights across multiple plant species. Here, we review the roles of 14-3-3s and AHAs in cell signaling to enhance plant responses to diverse environmental stresses. We analyzed the molecular evolution of key proteins that are associated with 14-3-3s and AHAs in plant growth and hormone signaling. The results revealed evolution, duplication, contraction, and expansion of 14-3-3s and AHAs in green plants. We also discussed the stress-specific expression of those 14-3-3s and AHAs in a eudicot (Arabidopsis thaliana), a monocot (Hordeum vulgare) and a moss (Physcomitrium patens) under abiotic stresses. We propose that 14-3-3s and H+-ATPases respond to abiotic stresses through many important targets and signaling components of phytohormones, which could be promising to improve plant tolerance to single or multiple environmental stresses. HighlightWe review the response and adaptation of 14-3-3s and AHAs to diverse environmental stimuli and we analyze the evolutionary features and molecular functions of 14-3-3s and AHAs.

plant biology↗

bioRxiv2023

Osseointegration of functionally-graded Ti6Al4V porous implants: Histology of the pore network

The additive manufacturing of titanium into porous geometries offers a means to generate low-stiffness endosseous implants with a greater surface area to improve osseointegration. In order to optimize pore size in the scaffolds, it is important to first understand the timeline of osseointegration in pre-clinical models. In this work, selective laser melting was used to produce gyroid-based scaffolds with a uniform pore size of 300 m or functionally-graded pore size from 600 m to 300 m before implantation in New Zealand white rabbit tibiae for 4 and 12 weeks. Initial in vitro assessment with Saos-2 cells showed favourable cell proliferation at pore sizes of 300 and 600 m. At four weeks, histological observations indicated some residual inflammation alongside neovessel infiltration into the scaffold interior and some early apposition of mineralized bone tissue. At twelve weeks, both scaffolds were filled with a mixture of adipocyte-rich marrow, micro-capillaries, and mineralized bone tissue. X-ray microcomputed tomography showed a higher bone volume fraction (BV/TV) and percentage of bone-implant contact (BIC) in the implants with 300 m pores than in the functionally-graded specimens, indicating that these smaller pore sizes may be favourable for osseointegration in leporine bone. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=77 SRC="FIGDIR/small/521963v1_ufig1.gif" ALT="Figure 1"> View larger version (26K): org.highwire.dtl.DTLVardef@2e319aorg.highwire.dtl.DTLVardef@baa9deorg.highwire.dtl.DTLVardef@e66075org.highwire.dtl.DTLVardef@c6c8c4_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioengineering↗

bioRxiv2022

Real-time denoising of fluorescence time-lapse imaging enables high-sensitivity observations of biological dynamics beyond the shot-noise limit

A fundamental challenge in fluorescence microscopy is the inherent photon shot noise caused by the inevitable stochasticity of photon detection. Noise increases measurement uncertainty, degrades image quality, and limits imaging resolution, speed, and sensitivity. To achieve high-sensitivity imaging beyond the shot-noise limit, we provide DeepCAD-RT, a versatile self-supervised method for effective noise suppression of fluorescence time-lapse imaging. We made comprehensive optimizations to reduce its data dependency, processing time, and memory consumption, finally allowing real-time processing on a two-photon microscope. High imaging signal-to-noise ratio (SNR) can be acquired with 10-fold fewer fluorescence photons. Meanwhile, the self-supervised superiority makes it a practical tool in fluorescence microscopy where ground-truth images for training are hard to obtain. We demonstrated the utility of DeepCAD-RT in extensive experiments, including in vivo calcium imaging of various model organisms (mouse, zebrafish larva, fruit fly), 3D migration of neutrophils after acute brain injury, and 3D dynamics of cortical ATP (adenosine 5-triphosphate) release. DeepCAD-RT will facilitate the morphological and functional interrogation of biological dynamics with minimal photon budget.

biophysics↗

bioRxiv2021

Loss of Protein Kinase D2 Activity Protects Against Bleomycin-induced Dermal Fibrosis in Mice

BackgroundDermal fibrosis occurs in many human diseases, particularly systemic sclerosis (SSc) where persistent inflammation leads to collagen deposition and fiber formation in skin and multiple organs. The family of protein kinase D (PKD) has been linked to inflammatory responses in various pathological conditions, however, its role in inflammation-induced dermal fibrosis has not been well defined. Here, using a murine fibrosis model that gives rise to dermal lesions similar to those in SSc, we investigated the role of PKD in dermal fibrosis in mice lacking PKD2 activity. MethodsHomozygous kinase-dead PKD2 knock-in mice (PKD2SSAA/SSAA-KI) were obtained through intercrossing mice heterozygous for PKD2S707A/S711A (PKD2SSAA). The wild-type and KI mice were subjected to repeated subcutaneous injection of bleomycin (BLM) to induce dermal inflammation and fibrosis. As controls, mice were injected with PBS. At the end of the experiment, mouse skin at the injection site was dissected, stained, and analyzed for morphological changes and expression of inflammatory and fibrotic markers. PKD-regulated signaling pathways were examined by real-time RT-qPCR and Western blotting. In a separate experiment, BLM-treated mice were administered with or without a PKD inhibitor, CRT0066101 (CRT). The effects of CRT on dermal fibrosis were analyzed similarly. The identity of the PKD expressing cells were probed using myeloid lineage markers CD45, CD68 in BLM-treated mouse tissues. ResultsDermal thickness and collagen fibers of kinase-dead PKD2-KI mice were significantly reduced in response to BLM treatment as compared to the wild-type mice. These mice also exhibited reduced -smooth muscle actin (-SMA) and collagen expression. At molecular levels, both transforming growth factor {beta}1 (TGF-{beta}1) and interleukin-6 (IL-6) mRNAs were decreased in the KI mice treated with BLM as compared to those in the wild-type mice. Similarly, CRT significantly blocked BLM-induced dermal fibrosis and inhibited the expression of -SMA, collagen, and IL-6 expression. Further analysis indicated that PKD2 was mainly expressed in CD45+/CD68+ myeloid cells that could be recruited to the lesional sites to promote the fibrotic process of the skin in response to BLM. ConclusionsKnock-in of the kinase-dead PKD2 or inhibition of PKD activity in mice protected against BLM-induced dermal fibrosis by reducing dermis thickness and expression of fibrotic biomarkers including -SMA, collagen, and inflammatory/fibrotic mediators including TGF-{beta}1 and IL-6. PKD2 does this potentially through modulating the recruitment and function of myeloid cells in skin of BLM-treated mice. Overall, our study demonstrated a potential critical role of PKD catalytic activity in inflammation-induced dermal fibrosis.

genetics↗

bioRxiv2021

A serotonergic axon-cilium synapse drives nuclear signaling to maintain chromatin accessibility

Chemical synapses between axons and dendrites mediate much of the brains intercellular communication. Here we describe a new kind of synapse - the axo-ciliary synapse - between axons and primary cilia. By employing enhanced focused ion beam - scanning electron microscopy on samples with optimally preserved ultrastructure, we discovered synapses between the serotonergic axons arising from the brainstem, and the primary cilia of hippocampal CA1 pyramidal neurons. Functionally, these cilia are enriched in a ciliary-restricted serotonin receptor, 5-hydroxytryptamine receptor 6 (HTR6), whose mutation is associated with learning and memory defects. Using a newly developed cilia-targeted serotonin sensor, we show that optogenetic stimulation of serotonergic axons results in serotonin release onto cilia. Ciliary HTR6 stimulation activates a non-canonical Gq/11-RhoA pathway. Ablation of this pathway results in nuclear actin and chromatin accessibility changes in CA1 pyramidal neurons. Axo-ciliary synapses serve as a distinct mechanism for neuromodulators to program neuron transcription through privileged access to the nuclear compartment.

cell biology↗

bioRxiv2021

Occurrence of COVID-19 symptoms during SARS-CoV-2 infection defines waning of humoral immunity

Approximately half of the SARS-CoV-2 infections occur without apparent symptoms, raising questions regarding long-term humoral immunity in asymptomatic individuals. Plasma levels of immunoglobulin G (IgG) and M (IgM) against the viral spike or nucleoprotein were determined for 25,091 individuals enrolled in a surveillance program in Wuhan, China. We compared 405 asymptomatic individuals with 459 symptomatic COVID-19 patients. The well-defined duration of the SARS-CoV-2 endemic in Wuhan allowed a side-by-side comparison of antibody responses following symptomatic and asymptomatic infections without subsequent antigen re-exposure. IgM responses rapidly declined in both groups. However, both the prevalence and durability of IgG responses and neutralizing capacities correlated positively with symptoms. Regardless of sex, age, and body weight, asymptomatic individuals lost their SARS-CoV-2-specific IgG antibodies more often and rapidly than symptomatic patients. These findings have important implications for immunity and favour immunization programs including individuals after asymptomatic infections. One-Sentence SummaryPrevalence and durability of SARS-CoV-2-specific IgG responses and neutralizing capacities correlate with COVID-19 symptoms.

immunology↗

bioRxiv2020

Performance and efficiency of machine learning algorithms for analyzing rectangular biomedical data

Most of the biomedical datasets, including those of omics, population studies and surveys, are rectangular in shape and have few missing data. Recently, their sample sizes have grown significantly. Rigorous analyses on these large datasets demand considerably more efficient and more accurate algorithms. Machine learning (ML) algorithms have been used to classify outcomes in biomedical datasets, including random forests (RF), decision tree (DT), artificial neural networks (ANN) and support vector machine (SVM). However, their performance and efficiency in classifying multi-category outcomes in rectangular data are poorly understood. Therefore, we aimed to compare these metrics among the 4 ML algorithms. As an example, we created a large rectangular dataset using the female breast cancers in the Surveillance, Epidemiology, and End Results-18 (SEER-18) database which were diagnosed in 2004 and followed up until December 2016. The outcome was the 6-category cause of death, namely alive, non-breast cancer, breast cancer, cardiovascular disease, infection and other cause. We included 58 dichotomized features from ~53,000 patients. All analyses were performed using MatLab (version 2018a) and the 10-fold cross validation approach. The accuracy in classifying 6-category cause of death with DT, RF, ANN and SVM was 72.68%, 72.66%, 70.01% and 71.85%, respectively. Based on the information entropy and information gain of feature values, we optimized dimension reduction (i.e. reduce the number of features in models). We found 22 or more features were required to maintain the similar accuracy, while the running time decreased from 440s for 58 features to 90s for 22 features in RF, from 70s to 40s in ANN and from 440s to 80s in SVM. In summary, we here show that RF, DT, ANN and SVM had similar accuracy for classifying multi-category outcomes in this large rectangular dataset. Dimension reduction based on information gain will significantly increase models efficiency while maintaining classification accuracy.

bioinformatics↗