Biology subjects

Nguyen, M.

Publications and source records attributed to Nguyen, M..

At least 91 records · Page 5Linked to original sources

bioRxiv2020

Predicting Antimicrobial Resistance Using Conserved Genes

A growing number of studies have shown that machine learning algorithms can be used to accurately predict antimicrobial resistance (AMR) phenotypes from bacterial sequence data. In these studies, models are typically trained using input features derived from comprehensive sets of known AMR genes or whole genome sequences. However, it can be difficult to determine whether genomes and their corresponding sets of AMR genes are complete when sequencing contaminated or metagenomic samples. In this study, we explore the possibility of using incomplete genome sequence data to predict AMR phenotypes. Machine learning models were built from randomly-selected sets of core genes that are held in common among the members of a species, and the AMR-conferring genes were removed based on their protein annotations. For Klebsiella pneumoniae, Mycobacterium tuberculosis, Salmonella enterica, and Staphylococcus aureus, we report that it is possible to classify susceptible and resistant phenotypes with average F1 scores ranging from 0.80-0.89 with as few as 100 conserved non-AMR genes, with very major error rates ranging from 0.11-0.23 and major error rates ranging from 0.10-0.20. Models built from core genes have predictive power in the cases where the primary AMR mechanism results from SNPs or horizontal gene transfer. By randomly sampling non-overlapping sets of core genes for use in these models, we show that F1 scores and error rates are stable and have little variance between replicates. Potential biases from strain-specific SNPs, phylogenetic sampling, and imbalances in the phylogenetic distribution of susceptible and resistant strains do not appear to have an impact on this result. Although these small core gene models have lower accuracies and higher error rates than models built from the corresponding assembled genomes, the results suggest that sufficient variation exists in the core non-AMR genes of a species for predicting AMR phenotypes. Overall this study suggests that building models from conserved genes may be a potentially useful strategy for predicting AMR phenotypes when genomes are incomplete.

bioinformatics↗

bioRxiv2020

The short-term effect of residential home energy retrofits on indoor air quality and microbial exposure: a case-control study

Weatherization of residential homes is a popular retrofit procedure to improve the energy efficiency of older homes by reducing building leakage. It is a vital tool in the fight against climate change. Several studies have evaluated the effect of weatherization on indoor pollutants such as formaldehyde and radon, but few studies have evaluated the effect of weatherization on indoor particulates and microbial exposure. In this study, we compared the effect of change in building leakage on indoor pollutants and bacterial communities in weatherized compared to non-weatherized single-family residential homes in New York State. Nine weatherized and eleven non-weatherized single-family homes in Tompkins County, New York were sampled twice: before and after the weatherization procedures for case homes, and at least 3 months apart for control homes that were not weatherized. We found a significant increase in both indoor-outdoor temperature ratio and living-area- and basement-radon levels of weatherized homes compared to control homes. The indoor-outdoor relative humidity ratio significantly decreased in weatherized compared to control homes. The indoor microbiome also became less similar to the outdoor community after weatherization. Compared to the changes in ventilation rate, temperature, relative humidity, and occupancy, the change in season was a more predictive measure of indoor bacterial concentration. Ventilation rate reduction from weatherization procedures led to an increase in indoor radon levels, as well as a warmer and less humid indoor environment. However, it did not affect indoor particulate mass concentration or indoor airborne bacteria load, and did only marginally affect the microbiome composition of residential homes. Finally, we found that changes in airborne bacterial load are more sensitive to shifts in season, whereas radon levels are more sensitive to ventilation rate.

microbiology↗

bioRxiv2020

Are bivalves susceptible to domestication selection? Using starvation tolerance to test for potential trait changes in eastern oyster larvae

Conservation efforts are increasingly being challenged by a rapidly changing environment, and for some aquatic species the use of captive rearing or selective breeding is an attractive option. However, captivity itself can impose unintended artificial selection known as domestication selection (adaptation to culture conditions). For most marine species, it is not known to what degree domestication selection affects traits related to fitness in the wild. To test for domestication selection in a marine bivalve, we focused on a fitness-related trait (larval starvation resistance) that could be altered under artificial selection. Using larvae produced from a wild population of Crassostrea virginica and a selectively bred, disease-resistant line we measured growth and survival during starvation versus standard algal diet (control) conditions. Larvae from both lineages showed a remarkable resilience to food limitation, possibly mediated by an ability to uptake and utilize dissolved organic matter for somatic maintenance. Water chemistry analysis showed dissolved organic carbon in filtered tank water to be at concentrations similar to natural river water. We observed that survival in larvae produced from the aquaculture line was significantly lower compared to larvae produced from wild broodstock (8 {+/-} 3% and 21 {+/-} 2%, respectively) near the end of a 10-day period with no food (phytoplankton). All larval cohorts had arrested growth during the starvation period and took at least two days to recover once food was reintroduced before resuming growth. Phenotypic differences between the wild and aquaculture lines suggest potential differences in the capacity to sustain extended food limitation, but this work requires replication with multiple selection lines and wild populations to make more general inferences about domestication selection. With this contribution we explore the potential for domestication selection in bivalves, discuss the physiological and fitness implications of reduced starvation tolerance, and aim to inspire further research on the topic.

evolutionary biology↗

bioRxiv2019

Tetrahymena Poc5 is a transient basal body component that is important for basal body maturation

Basal bodies (BBs) are microtubule-based organelles that template and stabilize cilia at the cell surface. Centrins ubiquitously associate with BBs and function in BB assembly, maturation, and stability. Human POC5 (hPOC5) is a highly conserved centrin-binding protein that binds centrins through Sfi1p-like repeats and is required for building full-length, mature centrioles. Here, we use the BB-rich cytoskeleton of Tetrahymena thermophila to characterize Poc5 BB functions. Tetrahymena Poc5 (TtPoc5) uniquely incorporates into assembling BBs and is then removed from mature BBs prior to ciliogenesis. Complete genomic knockout of TtPOC5 leads to a significantly increased production of BBs yet a markedly reduced ciliary density, both of which are rescued by reintroduction of TtPoc5. A second Tetrahymena POC5-like gene, SFR1, is similarly implicated in modulating BB production. When TtPOC5 and SFR1 are co-deleted, cell viability is compromised, and levels of BB overproduction are exacerbated. Overproduced BBs display defective transition zone formation and a diminished capacity for ciliogenesis. This study uncovers a requirement for Poc5 in building mature BBs, providing a possible functional link between hPOC5 mutations and impaired cilia.\n\nSUMMARY STATEMENTLoss of Tetrahymena thermophila Poc5 reveals an important role for this centrin-binding protein in basal body maturation, which also impacts basal body production and ciliogenesis.

cell biology↗

bioRxiv2019

An engineered CRISPR/Cas9 mouse line for simultaneous readout of lineage histories and gene expression profiles in single cells

Tracing the lineage history of cells is key to answering diverse and fundamental questions in biology. Particularly in the context of stem cell biology, analysis of single cell lineages in their native state has elucidated novel fates and highlighted heterogeneity of function. Coupling of such ancestry information with other molecular readouts represents an important goal in the field. Here, we describe the CARLIN (for CRISPR Array Repair LINeage tracing) mouse line and corresponding analysis tools that can be used to simultaneously interrogate the lineage and transcriptomic information of single cells in vivo. This model exploits CRISPR technology to generate up to 44,000 transcribed barcodes in an inducible fashion at any point during development or adulthood, is compatible with sequential barcoding, and is fully genetically defined. We have used CARLIN to identify intrinsic biases in the activity of fetal liver hematopoietic stem cell (HSC) clones and to uncover a previously unappreciated clonal bottleneck in the response of HSCs to injury. CARLIN also allows the unbiased identification of transcriptional signatures based on in vivo stem cell function without a need for markers or cell sorting.

developmental biology↗

bioRxiv2019

Predicting Alzheimer’s disease progression using deep recurrent neural networks

Early identification of individuals at risk of developing Alzheimers disease (AD) dementia is important for developing disease-modifying therapies. In this study, given multimodal AD markers and clinical diagnosis of an individual from one or more timepoints, we seek to predict the clinical diagnosis, cognition and ventricular volume of the individual for every month (indefinitely) into the future. We proposed and applied a minimal recurrent neural network (minimalRNN) model to data from The Alzheimers Disease Prediction Of Longitudinal Evolution (TADPOLE) challenge, comprising longitudinal data of 1677 participants (Marinescu et al. 2018) from the Alzheimers Disease Neuroimaging Initiative (ADNI). We compared the performance of the minimalRNN model and four baseline algorithms up to 6 years into the future. Most previous work on predicting AD progression ignore the issue of missing data, which is a prevalent issue in longitudinal data. Here, we explored three different strategies to handle missing data. Two of the strategies treated the missing data as a "preprocessing" issue, by imputing the missing data using the previous timepoint ("forward filling") or linear interpolation ("linear filling). The third strategy utilized the minimalRNN model itself to fill in the missing data both during training and testing ("model filling"). Our analyses suggest that the minimalRNN with "model filling" compared favorably with baseline algorithms, including support vector machine/regression, linear state space (LSS) model, and long short-term memory (LSTM) model. Importantly, although the training procedure utilized longitudinal data, we found that the trained minimalRNN model exhibited similar performance, when using only 1 input timepoint or 4 input timepoints, suggesting that our approach might work well with just cross-sectional data. An earlier version of our approach was ranked 5th (out of 53 entries) in the TADPOLE challenge in 2019. The current approach is ranked 2nd out of 63 entries as of June 3rd, 2020.

neuroscience↗

bioRxiv2019

Enhancer-dependence of gene expression increases with developmental age

How overall principles of gene regulation (the \"logic\") may change during ontogeny is largely unexplored. We compared transcriptomic, epigenomic and topological profiles in embryonic (EryP) and adult (EryD) erythroblasts. Despite reduced chromatin accessibility compared to EryP, distal chromatin of EryD is enriched in H3K27ac, Gata1 and Myb occupancy. In contrast to EryP-specific genes, which exhibit promoter-centric regulation through Gata1, EryD-specific genes employ distal enhancers for long-range regulation through enhancer-promoter looping, confirmed by Gata1 HiChIP. Genome editing demonstrated distal enhancers are required for gene expression in EryD but not in EryP. Applying a metric for enhancer-dependence of transcription, we observed a progressive reliance on enhancer control with increasing age of ontogeny among diverse primary cells and tissues of mouse and human origin. Our findings highlight fundamental and conserved differences in regulatory logic at distinct developmental stages, characterized by simpler promoter-centric regulation in embryonic cells and combinatorial enhancer-driven control in adult cells.\n\nHighlightsO_LIRegulation of embryonic-specific erythroid genes is promoter-centric through Gata1\nC_LIO_LIAdult-specific control is combinatorial enhancer-driven and requires Myb\nC_LIO_LIAdult specific genes have increased enhancer-promoter chromatin interactions\nC_LIO_LIEnhancer-dependence increases progressively with increasing developmental age\nC_LI

molecular biology↗

bioRxiv2019

Ventriculo-arterial uncoupling is associated with VO2 dependency in cardiac surgical patients

BackgroundThe clinical relevance of V-A (un)coupling in critically ill patients is under investigation. In this study we measured the association between V-A coupling and oxygen consumption (VO2) response in patients with acute circulatory instability following cardiac surgery.\n\nMethods and resultsSixty-one cardio-thoracic ICU patients who received fluid challenge or norepinephrine infusion were included. Arterial pressure, cardiac output (CO), heart rate (HR), arterial (EA), and ventricular elastances (EV), total indexed peripheral resistance (TPRi) were assessed before and after hemodynamic interventions. VO2 responders were defined as VO2 increase > 15 %. V-A coupling was evaluated by the ratio EA/EV. Left ventricle stroke work (SW) to pressure volume area (PVA) ratio was calculated. In the overall population, 24 patients (39%) were VO2 responders and 48 patients were uncoupled (i.e., EA/EV ratio > 1.3): 1.9 (1.6-2.4). Most of the uncoupled patients were classified as VO2 responders (28 of 31 patients, p=0.031). Changes in VO2 were correlated with those of TPRi, EA, EA/EV and CO. EA/EV ratio predicted VO2 increase with an AUC of 0.76 [95 % CI: 0.62-0.87]; p=0.001. In multivariate and principal component analyses, EA/EV and SW/PVA ratios were independently associated (P < 0.05) with VO2 response following interventions.\n\nConclusionsVO2 responders were characterized by baseline V-A uncoupling due to high EA and low EV. Baseline EA/EV and SW/PVA ratios were associated with VO2 changes independently of the hemodynamic intervention used. These results further underline the pathophysiological significance of V-A uncoupling in patients with hemodynamic instability.

physiology↗

bioRxiv2019

Model ensembles with different response variables for base and meta models: malaria disaggregation regression combining prevalence and incidence data

Maps of infection risk are a vital tool for the elimination of malaria. Routine surveillance data of malaria case counts, often aggregated over administrative regions, is becoming more widely available and can better measure low malaria risk than prevalence surveys. However, aggregation of case counts over large, heterogeneous areas means that these data are often underpowered for learning relationships between the environment and malaria risk. A model that combines point surveys and aggregated surveillance data could have the benefits of both but must be able to account for the fact that these two data types are different malariometric units. Here, we train multiple machine learning models on point surveys and then combine the predictions from these with a geostatistical disaggregation model that uses routine surveillance data. We find that, in tests using data from Colombia and Madagascar, using a disaggregation regression model to combine predictions from machine learning models trained on point surveys improves model accuracy relative to using the environmental covariates directly.

epidemiology↗

bioRxiv2018

Do Deep Neural Networks Outperform Kernel Regression for Functional Connectivity Prediction of Behavior?

There is significant interest in the development and application of deep neural networks (DNNs) to neuroimaging data. A growing literature suggests that DNNs outperform their classical counterparts in a variety of neuroimaging applications, yet there are few direct comparisons of relative utility. Here, we compared the performance of three DNN architectures and a classical machine learning algorithm (kernel regression) in predicting individual phenotypes from whole-brain resting-state functional connectivity (RSFC) patterns. One of the DNNs was a generic fully-connected feedforward neural network, while the other two DNNs were recently published approaches specifically designed to exploit the structure of connectome data. By using a combined sample of almost 10,000 participants from the Human Connectome Project (HCP) and UK Biobank, we showed that the three DNNs and kernel regression achieved similar performance across a wide range of behavioral and demographic measures. Furthermore, the generic feedforward neural network exhibited similar performance to the two state-of-the-art connectome-specific DNNs. When predicting fluid intelligence in the UK Biobank, performance of all algorithms dramatically improved when sample size increased from 100 to 1000 subjects. Improvement was smaller, but still significant, when sample size increased from 1000 to 5000 subjects. Importantly, kernel regression was competitive across all sample sizes. Overall, our study suggests that kernel regression is as effective as DNNs for RSFC-based behavioral prediction, while incurring significantly lower computational costs. Therefore, kernel regression might serve as a useful baseline algorithm for future studies.

neuroscience↗

bioRxiv2018

The origins and relatedness structure of mixed infections vary with local prevalence of P. falciparum malaria

Individual malaria infections can carry multiple strains of Plasmodium falciparum with varying levels of relatedness. Yet, how local epidemiology affects the properties of such mixed infections remains unclear. Here, we develop an enhanced method for strain deconvolution from genome sequencing data, which estimates the number of strains, their proportions, identity-by-descent (IBD) profiles and individual haplotypes. Applying it to the Pf3k data set, we find that the rate of mixed infection varies from 29% to 63% across countries and that 51% of mixed infections involve more than two strains. Further-more, we estimate that 47% of symptomatic dual infections contain sibling strains likely to have been co-transmitted from a single mosquito, and find evidence of mixed infections propagated over successive infection cycles. Finally, leveraging data from the Malaria Atlas Project, we find that prevalence correlates within Africa, but not Asia, with both the rate of mixed infection and the level of IBD.

epidemiology↗

bioRxiv2018

Using machine learning to predict antimicrobial minimum inhibitory concentrations and associated genomic features for nontyphoidal Salmonella

Nontyphoidal Salmonella species are the leading bacterial cause of food-borne disease in the United States. Whole genome sequences and paired antimicrobial susceptibility data are available for Salmonella strains because of surveillance efforts from public health agencies. In this study, a collection of 5,278 nontyphoidal Salmonella genomes, collected over 15 years in the United States, were used to generate XGBoost-based machine learning models for predicting minimum inhibitory concentrations (MICs) for 15 antibiotics. The MIC prediction models have average accuracies between 95-96% within {+/-} 1 two-fold dilution factor and can predict MICs with no a priori information about the underlying gene content or resistance phenotypes of the strains. By selecting diverse genomes for training sets, we show that highly accurate MIC prediction models can be generated with fewer than 500 genomes. We also show that our approach for predicting MICs is stable over time despite annual fluctuations in antimicrobial resistance gene content in the sampled genomes. Finally, using feature selection, we explore the important genomic regions identified by the models for predicting MICs. To date, this is one of the largest MIC modeling studies to be published. Our strategy for developing whole genome sequence-based models for surveillance and clinical diagnostics can be readily applied to other important human pathogens.

bioinformatics↗

bioRxiv2018

Brief Freezing Steps Lead to Robust Immunofluorescence in Drosophila Larval Brains

Drosophila melanogaster possess complex neuronal networks regulating sophisticated behavioral outputs that aid in studying the molecular mechanisms of neuronal function and neurodegenerative disease. Immunofluorescence (IF) techniques provide a way to visualize the spatiotemporal organization of these networks, permitting observation of their development, functional location, remodeling, and eventually - degradation. However, general immunostaining techniques do not always result in sufficient antibody penetration through the brain, and techniques used to enhance permeability can compromise structural integrity. We have found that freezing larval brains facilitates permeability with no apparent loss of antibody specificity or structural integrity. To demonstrate the advantage of this freezing technique, we compared results to two commonly used permeation methods: Detergent alone (Basic) and proteolytic degradation (Collagenase) techniques.\n\nSummaryHere we compare four different immunofluorescence techniques demonstrating that freezing Drosophila brains results in robust staining of small neurons in the larval brain without compromising structural integrity.

cell biology↗

bioRxiv2018

Impeding DNA Break Repair Enables Oocyte Quality Control

Oocyte quality control culls eggs with defects in meiosis. In mouse, oocyte death is triggered by defects in chromosome synapsis and recombination, which involve repair of programmed DNA double-strand breaks (DSBs) between homologous chromosomes. We show that RNF212, a SUMO ligase required for crossing over, also mediates oocyte quality control. Both physiological apoptosis and wholesale oocytes elimination in meiotic mutants require RNF212. RNF212 sensitizes cells to DSB-induced apoptosis within a narrow window when chromosomes desynapse during the transition into quiescence. Analysis of DNA damage during this transition implies that RNF212 impedes DSB repair. Consistently, RNF212 is required for HORMAD1, a negative regulator of inter-sister recombination, to associate with desynapsing chromosomes. We infer that oocytes impede repair of residual DSBs to retain a \"memory\" of meiotic defects that enables quality control processes. These results define the logic of oocyte quality control and suggest RNF212 variants may influence transmission of defective genomes.

genetics↗

bioRxiv2017

Shared understanding is correlated with shared neural responses in the default mode network

Humans have a striking ability to infer meaning from even the sparsest and most abstract forms of narratives. At the same time, flexibility in the form of a narrative is matched by inherent ambiguity in the interpretation of it. How does the brain represent subtle, idiosyncratic differences in the interpretation of abstract and ambiguous narratives? In this fMRI study, we scanned subjects watching a 7-min original animation that depicts a complex narrative through the movement of simple geometric shapes. We additionally scanned two separate groups listening to concrete verbal descriptions of either the social narrative or the physical properties of the movie. After scanning, all subjects freely recalled their interpretation of the stimulus. Using an intersubject representational similarity analysis, we compared the similarity of narrative interpretation across subjects, as measured using text analysis, with the similarity of neural responses, as measured using intersubject correlation (ISC). We found that the more similar two peoples interpretations of the abstract shape movie were, the more similar their neural responses were in the default mode network (DMN). Moreover, these shared responses were modality invariant: despite vast differences in stimulus properties, we found that the shapes movie and the verbal interpretation of the movie elicited shared responses in linguistic areas and a subset of the DMN when subjects shared interpretations. Together, these results suggest that the DMN is not only sensitive to subtle individual differences in narrative interpretation, but also resilient to large differences in the modality of the narrative.\n\nSignificance statementThe same narrative can be both communicated in different ways and interpreted in different ways. How are subtle, idiosyncratic differences in the interpretation of complex narratives presented in different forms represented in the brain? In this fMRI study, we show that the more similarly two people interpreted an ambiguous animation composed of moving shapes, the more similar their neural responses were in the Default Mode Network. In addition, by presenting the same narrative in a different form, we found shared responses across modalities when subjects shared interpretations despite the vast differences in modality of the stimuli. Together, these results suggest that the DMN is at once sensitive to individual differences in narrative interpretation and resilient to differences narrative modality.

neuroscience↗

bioRxiv2017

Developing an in silico minimum inhibitory concentration panel test for Klebsiella pneumoniae

Antimicrobial resistant infections are a serious public health threat worldwide. Whole genome sequencing approaches to apidly identify pathogens and predict antibiotic resistance phenotypes are becoming more feasible and may offer a way to reduce clinical test turnaround times compared to conventional culture-based methods, and in turn, improve patient outcomes. In this study, we use whole genome sequence data from 1668 clinical isolates of Klebsiella pneumoniae to develop a XGBoost-based machine learning model that accurately predicts minimum inhibitory concentrations (MICs) for 20 antibiotics. The overall accuracy of the model, within {+/-}1 two-fold dilution factor, is 92%. Individual accuracies are[≥]90% for 15/20 antibiotics. We show that the MICs predicted by the model correlate with known antimicrobial resistance genes. Importantly, the genome-wide approach described in this study offers a way to predict MICs for isolates without knowledge of the underlying gene content. This study shows that machine learning can be used to build a complete in silico MIC prediction panel for K. pneumoniae and provides a framework for building MIC prediction models for other pathogenic bacteria.

bioinformatics↗

bioRxiv2017

Multiplexing droplet-based single cell RNA-sequencing using natural genetic barcodes

Droplet-based single-cell RNA-sequencing (dscRNA-seq) has enabled rapid, massively parallel profiling of transcriptomes from tens of thousands of cells. Multiplexing samples for single cell capture and library preparation in dscRNA-seq would enable cost-effective designs of differential expression and genetic studies while avoiding technical batch effects, but its implementation remains challenging. Here, we introduce an in-silico algorithm demuxlet that harnesses natural genetic variation to discover the sample identity of each cell and identify droplets containing two cells. These capabilities enable multiplexed dscRNA-seq experiments where cells from unrelated individuals are pooled and captured at higher throughput than standard workflows. To demonstrate the performance of demuxlet, we sequenced 3 pools of peripheral blood mononuclear cells (PBMCs) from 8 lupus patients. Given genotyping data for each individual, demuxlet correctly recovered the sample identity of > 99% of singlets, and identified doublets at rates consistent with previous estimates. In PBMCs, we demonstrate the utility of multiplexed dscRNA-seq in two applications: characterizing cell type specificity and inter-individual variability of cytokine response from 8 lupus patients and mapping genetic variants associated with cell type specific gene expression from 23 donors. Demuxlet is fast, accurate, scalable and could be extended to other single cell datasets that incorporate natural or synthetic DNA barcodes.

bioinformatics↗

bioRxiv2017

The butterfly effect: amplification of local changes along the temporal processing hierarchy

Changing just a few words in a story can induce a substantial change in the overall narrative. How does the brain accumulate and process local and sparse changes, creating a unique situation model of the story, over the course of a real-life narrative? Recently, we mapped a hierarchy of processing timescales in the brain: from early sensory areas that integrate information over 10s-100s ms, to high-order areas that integrate information over many seconds to minutes. Based on this hierarchy, we hypothesize that early sensory areas would be sensitive to local changes in word use, but that there will be increasingly divergent neural responses along the processing hierarchy as higher-order areas accumulate and amplify these local changes. To test this hypothesis, we created two structurally related but interpretively distinct narratives by changing some individual words. We found that the neural response distance between the stories was amplified as story information is transferred from low-level regions (e.g. early auditory cortex) to high-level regions (e.g precuneus and prefrontal cortex) and that the neural difference between stories is highly correlated with an areas ability to integrate information over time. Our results suggest a neural mechanism by which two similar situations become easy to distinguish.

neuroscience↗