Biology subjects

Green, J.

Publications and source records attributed to Green, J..

At least 37 records · Page 2Linked to original sources

bioRxiv2022

Matrix fibroblast function during alveolarization is dependent on GATA6

Alveolarization is dependent on myo-, matrix- and lipo- fibroblast functions by interstitial PDGFRa+ fibroblasts. While these fibroblasts are derived from GLI and PDGFRa expressing fibroblasts, the transcriptional control of their functional specification remains unknown. Perinatally, the transcription factor GATA6 is upregulated in PDGFRa+ fibroblasts. To study the role of GATA6 during fibroblast differentiation, we generated PDGFRaCreER/GATA6flx/flx mice and deleted GATA6 in the perinatal period and in adult mice prior to left lobe pneumonectomy. Loss of GATA6 in the PDGFRa+-fibroblasts impaired alveolarization, and extracellular matrix deposition, in association with increased TCF21 expression and lipofibroblast differentiation. Loss of GATA6 in PDGFRa+ fibroblasts resulted in loss of alveolar type 1 (AT1) cells and gain of transitional alveolar type 2 (AT2) cells. Loss of GATA6 was associated with reduced WNT signaling. Restoration of WNT signaling in GATA6 deficient alveolar lung organoids restored AT2 and AT1 cell differentiation. GATA6 induces matrix fibroblast functions and represses lipofibroblast functions, serving as key regulator of fibroblast differentiation during alveolarization and regeneration. Present findings link matrix fibroblast functions with the ability of transitional AT2 cells to differentiate into AT1 cells. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=135 SRC="FIGDIR/small/494950v1_ufig1.gif" ALT="Figure 1"> View larger version (20K): org.highwire.dtl.DTLVardef@47375forg.highwire.dtl.DTLVardef@184630borg.highwire.dtl.DTLVardef@8285b3org.highwire.dtl.DTLVardef@117e47a_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGraphical abstract:C_FLOATNO C_FIG

developmental biology↗

bioRxiv2022

3D Visualisation of trans-syncytial nanopores provides a pathway for paracellular diffusion across the human placental syncytiotrophoblast

The placental syncytiotrophoblast, a syncytium without cell-cell junctions, is the primary barrier between the mother and the fetus. Despite no apparent anatomical pathway for paracellular diffusion of solutes across the syncytiotrophoblast size-dependent paracellular diffusion is observed. Here we report data demonstrating that the syncytiotrophoblast is punctuated by trans-syncytial nanopores (TSNs). These membrane-bound TSNs directly connect the maternal and fetal facing sides of the syncytiotrophoblast, providing a pathway for paracellular diffusion between the mother and fetus. Mathematical modelling of TSN permeability based on their 3D geometry suggests that 10-37 million TSNs per cm3 of placental tissue could explain experimentally observed placental paracellular diffusion. TSNs may mediate physiological hydrostatic and osmotic pressure homeostasis between the maternal and fetal circulations but also expose the fetus to pharmaceuticals, environmental pollutants and nanoparticles.

physiology↗

bioRxiv2022

Arresting microbiome development limits immune system maturation and resistance to infection.

Disruptions to the intestinal microbiome during weaning lead to long-term negative effects on host immune function. However, the critical host-microbe interactions occurring during weaning required for healthy immune system development remain poorly understood. We find that restricting microbiome maturation during weaning leads to stunted immune system development and increased susceptibility to enteric infection. We developed a gnotobiotic mouse model of the early-life microbiome designated as Pediatric Community (PedsCom). This nine-member consortium of microbes derived from intestinal microbiomes of preweaning mice stably colonized germfree adult mice and was efficiently transmitted to offspring for multiple generations. Unexpectedly, the relative abundance of PedsCom microbes were largely unaffected by the transition from a milk-based to a fiber rich solid food diet. PedsCom mice developed less peripheral regulatory T cells and Immunoglobulin A, hallmarks of microbiota-driven immune system development. Consistent with defects in maturation, adult PedsCom mice retain high susceptibility to salmonella infection characteristic of young mice and humans. Altogether, our work illustrates how the post-weaning transition in intestinal microbiome composition contributes to normal immune maturation and protection from enteric infection. Accurate modelling of the pre-weaning microbiome provides a window into the microbial requirements of healthy immune development and suggests an opportunity to design microbial interventions at weaning to improve immune system development in human infants. One Sentence SummaryArresting microbiome development stunts immune ontogeny

immunology↗

bioRxiv2021

Mining hidden knowledge: Embedding models of cause-effect relationships curated from the biomedical literature

We explore the use of literature-curated signed causal gene expression and gene-function relationships to construct un-supervised embeddings of genes, biological functions, and diseases. Our goal is to prioritize and predict activating and inhibiting functional associations of genes, and to discover hidden relationships between functions. As an application, we are particularly interested in the automatic construction of networks that capture relevant biology in a given disease context. We evaluated several unsupervised gene embedding models leveraging literature-curated signed causal gene expression findings. Using linear regression, it is shown that, based on these gene embeddings, gene-function relationships can be predicted with about 95% precision for the highest scoring genes. Function embedding vectors, derived from parameters of the linear regression model, allow to infer relationships between different functions or diseases. We show for several diseases that gene and function embeddings can be used to recover key drivers of pathogenesis, as well as underlying cellular and physiological processes. These results are presented as disease-centric networks of genes and functions. To illustrate the applicability of the computed gene and function embeddings to other machine learning tasks we expanded the embedding approach to drug molecules, and used a simple neural network to predict drug-disease associations.

bioinformatics↗

bioRxiv2021

Exposure to nitric oxide drives transition to resuscitation-promoting factor-dependency in mycobacteria

During infection Mycobacterium tuberculosis (Mtb) forms differentially culturable (DC) subpopulations that are recalcitrant to treatment and undetectable using standard diagnostic tools. DC Mtb are revealed in liquid media, their revival is often stimulated by resuscitation-promoting factors (Rpfs), secreted peptidoglycan-remodelling enzymes, and prevented by Rpf inhibitors. Here we investigated the role of nitric oxide (NO) in generation of Rpf- dependent DC Mtb, using murine macrophage infection models and treatment with a synthetic NO donor (NOD). Mtb subpopulations were assessed by colony-forming unit counting on agar or by limiting dilution Most Probable Number assays in liquid media with or without Rpf inhibitor. Rpf-dependent DC Mtb were detected following infection of interferon-{gamma} induced macrophages capable of producing NO, but not when iNOS was inactivated. NOD treatment also induced transition to the Rpf-dependent DC phenotype which was accompanied by global transcriptomic changes resulting in the dramatic down-regulation of rpfA-E gene expression. Furthermore, the DC phenotype was partially reverted by artificial over-expression of Rpfs. This study elucidates molecular mechanisms underlying the generation of DC Mtb, which are the dominant population recovered from clinical tuberculosis samples, with implications for improving both tuberculosis diagnostics and treatments.

microbiology↗

bioRxiv2021

NLRP3 activation in response to disrupted endocytic traffic

Inflammation driven by the NLRP3 inflammasome is coordinated through multiple signaling pathways and with a poorly defined regulation by sub-cellular organelles. Here, we tested the hypothesis that NLRP3 senses disrupted endosome trafficking to trigger inflammasome formation and inflammatory cytokine secretion. NLRP3-activating stimuli disrupted endosome trafficking and triggered localization of NLRP3 to vesicles positive for endosome markers and the inositol lipid PtdIns4P. Chemical disruption of endosome trafficking sensitized macrophages to the NLRP3 activator imiquimod driving enhanced inflammasome activation and cytokine secretion. Together these data suggest that NLRP3 is capable of sensing disruptions in the trafficking of endosomal cargoes, and that this may explain in part the spatial activation of the NLRP3 inflammasome complex. These data highlight new mechanisms amenable for the therapeutic targeting of NLRP3. One-Sentence SummaryNLRP3 senses disruptions in endosome trafficking to trigger the formation of an inflammasome complex and initiate an inflammatory response.

immunology↗

bioRxiv2021

The Telomere Length Landscape of Localized Prostate Cancer

Replicative immortality is a hallmark of cancer, and can be achieved through telomere lengthening and maintenance. We report telomere lengths (TLs) of 392 localized prostate cancer tumours and characterize their relationship to genomic, transcriptomic and proteomic features. Shorter tumour TLs were associated with elevated genomic instability, including single-nucleotide variants, indels and structural variants. Genes involved in cell proliferation and signaling were correlated with tumour TL at all levels of the central dogma. TL was also associated with multiple clinical features of a tumour. Longer TLs in non-tumour samples were associated with a lower rate of biochemical relapse after definitive local therapy. Our analysis integrates multi-omics data to illuminate the relationship of specific genomic alterations in a tumour and TL in prostate cancer. Although the role of telomere length in cancer has been well studied, its association to genomic features is less well known. We describe the multi-level integration of telomere length, genomics, transcriptomics and proteomics in localized prostate cancer. Patient Summary We examined the association between telomere length and multiple omics-level data in prostate cancer. We observed that traditional telomere mutations are rare in prostate cancer and that telomere length is associated with multiple measure of genomic instability.

genomics↗

bioRxiv2020

Dietary folic acid deficiency impacts hippocampal morphology and cortical acetylcholine metabolism in adult male and female mice

ObjectiveOne-carbon metabolism is a metabolic network that integrates nutritional signals with biosynthesis, redox homeostasis, and epigenetics. There are sex differences in hepatic one-carbon metabolism. However, it is unclear whether there are sex differences in dietary deficiencies of one-carbon metabolism in the brain. The aim of this study was to investigate if sex modulates the effects of dietary folic acid deficiency in brain tissue using a mouse model. MethodsMale and female C57Bl/6J mice were placed on a folic acid deficient (FD) or control diet (CD) at six weeks of age. Mice were maintained on these diets for six months, after which animals were euthanized and brain tissue and serum were collected for analysis. Serum folate levels were measured. In brain tissue, hippocampal volume and morphology including Cornu Ammonis 1 and 3 (CA1; CA3), and dentate gyrus thickness were measured. Apoptosis within the hippocampus was assessed using active caspase-3 immunofluorescence staining. Additionally, cortical acetylcholine metabolism was measured in brain tissue using immunofluorescence staining of acetylcholinesterase (AChE), or choline acetyltransferase (ChAT), and neuronal nuclei (NeuN). ResultsMale and female FD mice had reduced serum levels of folate. Both males and females maintained on a FD showed a decrease in the thickness of the hippocampal CA1-CA3 region. Interestingly, there was a sex difference in the levels of active caspase-3 within the CA3 region of the hippocampus. In cortical tissue, there were increased levels of neuronal ChAT and reduced levels of AChE in FD females and male mice. ConclusionsThe results indicated that FD impacts hippocampal morphology and cortical neuronal acetylcholine metabolism. The data from our study indicate that there was only one sex difference and that was in hippocampal apoptosis. Our study provides little evidence that sex modulates the effects of dietary folate deficiency on hippocampal morphology and cortical neuronal acetylcholine metabolism.

neuroscience↗

bioRxiv2020

Cytomegaloviral determinants of CD8+ T cell programming and RhCMV/SIV vaccine efficacy

Simian immunodeficiency virus (SIV) insert-expressing, 68-1 Rhesus Cytomegalovirus (RhCMV/SIV) vectors elicit major histocompatibility complex (MHC)-E- and -II-restricted, SIV-specific CD8+ T cell responses, but the basis of these unconventional responses and their contribution to demonstrated vaccine efficacy against SIV challenge in the rhesus monkeys (RMs) has not been characterized. We demonstrate that these unconventional responses resulted from a chance genetic rearrangement in 68-1 RhCMV that abrogated the function of eight distinct immunomodulatory gene products encoded in two RhCMV genomic regions (Rh157.5/.4 and Rh158-161). Differential repair of these genes with either RhCMV-derived or orthologous human CMV (HCMV)-derived sequences (UL128/130; UL146/147) leads to either of two distinct CD8+ T cell response types - MHC-Ia-restricted-only, or a mix of MHC-II- and MHC-Ia-restricted CD8+ T cells. Despite response magnitude and functional differentiation being similar to RhCMV 68-1, neither alternative response type mediated protection against SIV challenge. These findings implicate MHC-E-restricted CD8+ T cell responses as mediators of anti-SIV efficacy and indicate that translation of RhCMV/SIV vector efficacy to humans will likely require deletion of all the genes that inhibit these responses from the HCMV/HIV vector. One-sentence summaryEight genes in two spatially distinct RhCMV gene regions control induction of unconventionally restricted CD8+ T cell responses and the efficacy of RhCMV/SIV vaccine vectors against SIV challenge.

immunology↗

bioRxiv2020

Mycobacterium smegmatis expands across surfaces using hydraulic sliding

Mycobacterium smegmatis spreads over soft agar surfaces by sliding motility, a form of passive motility in which growth and reduction of surface adhesion enable the bacteria to push each other outwards. Hence, sliding motility is mostly associated with round colonies. However, M. smegmatis sliding colonies can also produce long, pointed dendrites. Round sliding colonies were readily reproduced, but our non-round colonies were different from those seen previously. The latter (named digitate colonies) had centimetre-long linear protrusions, containing a central channel filled with a free-flowing suspension of M. smegmatis and solid aggregates. Digitate colonies had both a surface pellicle and an inner biofilm component surrounding a central channel, which sat in a cleft in the agar. Time-lapse microscopy showed that the expansion of the fluid-filled channel enabled the lengthwise extension of the protrusions without any perceptible growth of the bacteria taking place. These observations represent a novel type of sliding motility, named hydraulic sliding, associated with a specialised colony structure and the apparent generation of force by expansion of a liquid core. As this structure requires pellicle formation without an initial liquid culture it implies the presence of an unstudied mycobacterial behaviour that may be important for colonisation and virulence. Originality-Significance StatementThis study is the first to identify a new form of passive motility in the mycobacteria; hydraulic sliding, in which liquid expansion is the cause of motility. This form of motility has so far never been described in bacteria. The study also reveals new ways mycobacteria can form biofilms and colonize complex three-dimensional substrates, aspects of mycobacterial biology that are important for infection, pathogenesis and vaccine development. Author SummaryMycobacterium smegmatis is used as a non-pathogenic model organism for pathogenic mycobacteria. During growth, M. smegmatis can move passively over soft agar surfaces by a process called sliding motility, whereby colony growth directly pushes cells outwards. Although passive, sliding motility is believed to be important in allowing bacteria to colonise surfaces. Sliding motility however does not fully account for how M. smegmatis produces dendritic colonies. We attempted to generate dendritic colonies but found instead that the cells produced colonies that had larger protrusions radiating from them (digitate colonies). Digitate colonies are a previously unobserved phenomenon, in that the bacteria create a biofilm-lined, fluid-filled, pellicle-covered, deep cleft in the agar and move across the surface by the expansion of the contained liquid core of the protrusions. Given the new structure and the new mechanism of expansion we have termed this set of behaviours hydraulic sliding. These observations are important as it is a new variation in the way bacteria can move, generate biofilms (notably mycobacterial pellicle) and colonize complex three-dimensional substrates.

microbiology↗

bioRxiv2020

The Coronavirus Network Explorer: Mining a large-scale knowledge graph for effects of SARS-CoV-2 on host cell function

Building on recent work that identified human host proteins that interact with SARS-CoV-2 viral proteins in the context of an affinity-purification mass spectrometry screen, we use a machine learning-based approach to connect the viral proteins to relevant biological functions and diseases in a large-scale knowledge graph derived from the biomedical literature. Our aim is to explore how SARS-CoV-2 could interfere with various host cell functions, and also to identify additional drug targets amongst the host genes that could potentially be modulated against COVID-19. Results are presented in the form of interactive network visualizations, that allow exploration of underlying experimental evidence. A selection of networks is discussed in the context of recent clinical observations.

systems biology↗

bioRxiv2020

Pretreatment of aged mice with retinoic acid restores alveolar regeneration via upregulation of reciprocal PDGFRA signaling

ObjectivesIdiopathic Pulmonary Fibrosis (IPF) primarily affects the aged population and is characterized by failure of alveolar regeneration leading to loss of alveolar type 1 cells (AT1). Aged mouse models of lung repair have demonstrated that regeneration fails with increased age. Mouse and rat lung repair models have shown retinoic acid (RA) treatment can restore alveolar regeneration. Herein we seek to determine the signaling mechanisms by which RA treatment prior to injury supports alveolar differentiation. DesignPartial pneumonectomy (PNX) lung injury model and next generation sequencing of sorted cell populations are used to uncover molecular targets regulating alveolar repair. In-vitro organoids generated from Mouse or IPF patient epithelial cells co-cultured with young, aged, or RA pretreated murine mesenchyme are used to test potential targets. Main outcome measurementsKnown alveolar epithelial cell differentiation markers, including HOPX and AGER for AT1 cells are used to assess outcome of treatments. ResultsGene expression analysis of sorted fibroblasts and epithelial cells isolated from lungs of young, aged, and RA treated aged mice predicted increased PDGFA signaling that coincided with regeneration and alveolar epithelial differentiation. Addition of PDGFA induced AT1 and AT2 alveolar differentiation in both mouse and human IPF lung organoids generated with aged fibroblasts and PDGFA monoclonal antibody blocked AT1 cell differentiation in organoids generated with young murine fibroblasts. ConclusionsOur data support the concept that reciprocal PDGFA signaling activates regenerative fibroblasts that support alveolar epithelial cell differentiation and repair, providing a potential therapeutic strategy to influence the pathogenesis of IPF. Key QuestionWhich epithelial-mesenchymal crosstalk pathways are activated by RA pretreatment of aged lungs that support realveolarization after partial pneumonectomy surgery? Bottom LineIncreased PDGFA/PDGFRA signaling in aged lungs promotes regenerative activation of interstitial matrixfibroblast which is required for AT2 to AT1 differentiation and alveolar regeneration. Read OnIn-vitro and in-vivo analysis demonstrated that PDGFA signaling supports alveolar matrixfibroblast and AT1 epithelial cell differentiation, both necessary for alveolar regeneration in aged lungs.

cell biology↗

bioRxiv2020

Pre-winter fattening and fat loss during wing moult: the annual cycle of fat deposition in captive barnacle geese (branta leucopsis)

Many different physiological changes have been observed in wild waterfowl during the flightless stage of wing moult, including a loss of body mass. Previously we established that captive barnacle geese (Branta leucopsis) underwent this characteristic decrease in body mass during their wing moult, even though they had unlimited and unrestricted access to food. In the present study we aimed to determine if this body mass loss during moult comprised mainly a reduction in fat stores, and to ascertain if the captive geese undergo pre-migratory and pre-winter fattening over a similar temporal scale to their wild conspecifics. The non-destructive technique of deuterium oxide isotope dilution was employed to provide repeated measurements of estimated fat deposition from a captive flock of fourteen barnacle geese. Birds were injected with deuterium oxide at 7 distinct intervals for one annual cycle. During the flightless period of the moult, body fat decreased by approximately 40% from the pre-moult value. During late-September and early October, body fat reached its highest point in the annual cycle, both as an absolute value and as a percentage of total body mass. We propose that while the energetic cost of wing moult is not the ultimate cause of fat loss in moulting barnacle geese, the approximate 212 g of fat catabolised during moult would provide sufficient energy to cover the cost of the replacement of the flight feathers, estimated to be 6384 kJ, over an approximate 42 day period. We conclude that the previously recorded increase in metabolism during moult in the geese, led to the use of endogenous fat reserves because the birds reduced rather than increased their food intake rates owing to the increased risk of predation when flightless. We also conclude that captive barnacle geese do undergo pre-winter and pre-migratory fattening, providing further evidence of the innate nature of these fat deposition cycles.Competing Interest StatementThe authors have declared no competing interest.View Full Text

zoology↗

bioRxiv2020

Mechanics Of Flight Feathers: Effects Of Captivity?

ABSTRACTFeathers act as aerodynamic cantilevers, and to withstand the prolonged cyclical loading that occurs during flight, feathers must be stiff, lightweight and strong. We experimentally tested the differences in feather structure, primarily stiffness and size, between (a) wild and captive Barnacle Geese Branta leucopsis, and (b) primary feathers dropped during the annual flight feather moult, and those feathers freshly regrown during the moult process. We found that, despite having undergone a 5,000km round-trip migration, flight feathers dropped during moult in the wild geese were stiffer than those measured in the captive geese, both for those dropped during moult and those re-grown. We propose that this may be related to diet or stress in the captive geese.Competing Interest StatementThe authors have declared no competing interest.View Full Text

zoology↗

bioRxiv2020

Biomarker-guided treatment strategies for ovarian cancer identified from a heterogeneous panel of patient-derived tumor xenografts

Advanced ovarian cancers are a leading cause of cancer-related death in women. Such cancers are currently treated with surgery and chemotherapy which is often temporarily successful but exhibits a high rate of relapse after which treatment options are few. Here we assess the responses of a panel of patient-derived ovarian cancer xenografts (PDXs) to 19 mono and combination therapies, including small molecules and antibody-drug conjugates. The PDX panel aimed to mimic the heterogeneity of disease observed in patients, and exhibited a distribution of responsiveness to standard of care chemotherapy similar to human clinical data. Three monotherapies and one drug combination were found to be active in different subsets of PDXs. By analyzing gene expression data we identified gene expression biomarkers predictive of responsiveness to each of three novel targeted therapy regimens. While no single treatment had as high a response rate as chemotherapy, nearly 90% of PDXs were eligible for and responded to at least one biomarker-guided treatment, including tumors resistant to standard chemotherapy. Biomarker frequency was similar in human patients, suggesting the possibility of a new therapeutic approach to ovarian cancer and demonstrating the potential power of PDX-based trials in broadening the reach of precision cancer medicine.

cancer biology↗

bioRxiv2020

Near-chromosome level genome assembly of the fruit pest Drosophila suzukii using long-read sequencing

Over the past decade, the spotted wing Drosophila, Drosophila suzukii, has invaded Europe and America and has become a major agricultural pest in these areas, thereby prompting intense research activities to better understand its biology. Two draft genome assemblies based on short-read sequencing were released in 2013 for this species. Although valuable, these resources contain pervasive assembly errors and are highly fragmented, two features limiting their values. Our purpose here was to improve the assembly of the D. suzukii genome. For this, we generated PacBio long-read sequencing data at 160X sequence coverage and assembled a novel, contiguous D. suzukii genome. We obtained a high-quality assembly of 270 Mb (with 546 contigs, a N50 of 2.6Mb, a L50 of 15, and a BUSCO score of 95%) that we called WT3-2.0. We found that despite 16 rounds of full-sib crossings the D. suzukii strain that we sequenced has maintained high levels of polymorphism in some regions of its genome (ca. 19Mb). As a consequence, the quality of the assembly of these regions was reduced. We explored possible origins of this high residual diversity, including the presence of structural variants and a possible heterogeneous admixture pattern of North American and Asian ancestry. Overall, our WT3-2.0 assembly provides a higher quality genomic resource compared to the previous one in terms of general assembly statistics, sequence quality and gene annotation. This new D. suzukii genome assembly is therefore an improved resource for high-throughput sequencing approaches, as well as manipulative genetic technologies to study D. suzukii.

ecology↗

bioRxiv2019

Quantitative assessment of fecal contamination in multiple environmental sample types in urban communities in Dhaka, Bangladesh using SaniPath microbial approach

Rapid urbanization has led to a growing sanitation crisis in urban areas of Bangladesh and potential exposure to fecal contamination in the urban environment due to inadequate sanitation and poor fecal sludge management. Limited data are available on environmental fecal contamination associated with different exposure pathways in urban Dhaka. We conducted a cross-sectional study to explore the magnitude of fecal contamination in the environment in low-income, high-income, and transient/floating neighborhoods in urban Dhaka. Ten samples were collected from each of 10 environmental compartments in 10 different neighborhoods (4 low-income, 4 high-income and 2 transient/floating neighborhoods). These 1,000 samples were analyzed with the IDEXX-Quanti-Tray technique to determine most-probable-number (MPN) of E. coli. Samples of open drains (6.91 log10 MPN/100 mL), surface water (5.28 log10 MPN/100 mL), floodwater (4.60 log10 MPN/100 mL), produce (3.19 log10 MPN/serving), soil (2.29 log10 MPN/gram), and street food (1.79 log10 MPN/gram) had the highest mean log10 E. coli contamination compared to other samples. The contamination concentrations did not differ between low-income and high-income neighborhoods for shared latrine swabs, open drains, municipal water, produce, and street foodsamples. E. coli contamination were significantly higher (p <0.05) in low-income neighborhoods compared to high-income for soil (0.91 log10 MPN/gram, 95% CI, 0.39, 1.43), bathing water (0.98 log10 MPN/100 mL, 95% CI, 0.41, 1.54), non-municipal water (0.64 log10 MPN/100 mL, 95% CI, 0.24, 1.04), surface water (1.92 log10 MPN/100 mL, 95% CI, 1.44, 2.40), and floodwater (0.48 log10 MPN/100 mL, 95% CI, 0.03, 0.92) samples. E. coli contamination were significantly higher (p<0.05) in low-income neighborhoods compared to transient/floating neighborhoods for drain water, bathing water, non-municipal water and surface water. Future studies should examine behavior that brings people into contact with the environment and assess the extent of exposure to fecal contamination in the environment through multiple pathways and associated risks.

microbiology↗

bioRxiv2019

Multi-view Co-training for MicroRNA Prediction

MicroRNA (miRNA) are short, non-coding RNAs involved in cell regulation at post-transcriptional and translational levels. Numerous computational predictors of miRNA been developed that generally classify miRNA based on either sequence- or expression-based features. While these methods are highly effective, they require large labelled training data sets, which are often not available for many species. Simultaneously, emerging high-throughput wet-lab experimental procedures are producing large unlabelled data sets of genomic sequence and RNA expression profiles. Existing methods use supervised machine learning and are therefore unable to leverage these unlabelled data. In this paper, we design and develop a multi-view co-training approach for the classification of miRNA to maximize the utility of unlabelled training data by taking advantage of multiple views of the problem. Starting with only 10 labelled training data, co-training is shown to significantly increase classification accuracy of both sequence- and expression-based classifiers, without requiring any new labelled training data. After 11 iterations of co-training, the expression-based view of miRNA classification experiences an average increase in AUPRC of 15.81% over six species, compared to 11.90% for self-training and 4.84% for passive learning. Similar results are observed for sequence-based classifiers with increases of 46.47%, 39.53% and 29.43%, for co-training, self-training, and passive learning, respectively. The final co-trained sequence and expression-based classifiers are integrated into a final confidence-based classifier which shows improved performance compared to each individual view. This study represents the first application of multi-view co-training to miRNA prediction and shows great promise, particularly for understudied species with few available training data.\n\nAvailabilityCode is available at https://github.com/GreenCUBIC/miRNA_MVCT. All datasets are publicly available with accession numbers listed in the manuscript.\n\nContactjrgreen@sce.carleton.ca

bioinformatics↗