Biology subjects

Ridenhour, B. J.

Publications and source records attributed to Ridenhour, B. J..

11 recordsLinked to original sources

bioRxiv2026

IQC: A Novel Criterion for Assessing Feature Selection Stability in High-Dimensional Analyses

In recent years, the issue of reproducibility in scientific studies has been at the forefront of scientific discussion. Many of these issues arise due to poor employment of statistical methods, making it crucial to address these knowledge gaps for frequently used methods. Elastic net regression (ENET) is one such algorithm used frequently in biology and public health due to its ability to select important predictors (also known as "features") from large data sets. However, ENET struggles with consistent feature selection and hyperparameter choice in these applications due to low observation counts and noisy data. This limitation hinders ENET's effectiveness in biological research due to low model reproducibility and predictor interpretability. We propose the Instability Quotient Criterion (IQC) as a metric to quantify the consistency of selected features for feature selection models such as ENET, with a specific emphasis on biological applications. IQC leverages ensemble modeling and Shannon's entropy in order to determine the stability of selected features derived from ENET models. We demonstrate IQC's robustness via simulation studies, showing that IQC is capable of detecting significant decreases in model stability with lower observation counts, higher dimensionality, and higher proportions of noisy signal. We additionally utilize simulation to provide a naive threshold for low versus high stability. Finally, we applied IQC to a case study to illustrate how our method can be integrated into a larger analysis pipeline when looking at microarray gene expression data. IQC indicated stability issues for differentially-expressed genes when classifying acute leukemia subtypes in spite of high classification success. Lastly, we discuss how model instability impedes interpretation of biologically significant genes for discerning cancer subtypes. Overall, we present IQC as a robust evaluation metric that can easily be incorporated into high-dimensional biological analyses.

bioinformatics↗

bioRxiv2026

An Aging Risk-Factor Scale: Biomarkers of Renal Disease and Anemia are Primary Predictors of Three-Year Survival in Common Marmosets (Callithrix jacchus)

Valid animal models are needed to evaluate how age-related changes in kidney function influence healthspan. Aging marmosets frequently develop renal insufficiency with anemia and exhibit reductions in body mass and metabolic rate. However, it remains unclear which age-related changes predict survival and which thresholds indicate increased mortality risk. We prospectively evaluated age, body composition, resting energy expenditure, hematology, and blood chemistry as predictors of 3-year survival in female and male marmosets (n = 66), 2-16 years of age. Objectives were to identify prognostic markers, define high-risk thresholds, and to develop and test a composite risk-factor scale for mortality screening in captivity. A 10-variable model showed the best predictive performance in multivariable Cox proportional hazards modeling, and was retained for further analysis (concordance = 0.881, p < 0.001). ROC curves using Youdens Index and AUC identified high-risk thresholds for predictors in the multivariable model, and threshold-defined categories were evaluated by Kaplan-Meier survival analysis. The 10 binary risk-factors were combined into a composite scale scored from 0 to 10 and tested with Cox regression. The scale explained approximately 42% of variance in survival and each additional risk factor increased mortality risk 1.75-fold (95% CI: 1.43-2.14, p < 0.001). Marmosets with [≥]7 risk factors exhibited a 19-month reduction in survival, and this high-risk threshold predicted 3-year survival with 89.4% accuracy. Results support the scale as a screening tool for mortality risk and highlight the high prevalence of age-associated renal disease and anemia in marmosets.

physiology↗

bioRxiv2025

Validation of Body Condition Scoring as a Screening Test for Low Body Condition and Obesity in Common Marmosets (Callithrix jacchus)

Assessing body weight is common practice for monitoring health in common marmosets (Callithrix jacchus). Body composition analysis via quantitative magnetic resonance (QMR) is a more in-depth assessment allowing measurements of lean and fat mass, but it is expensive and remains unavailable to most. Alternatively, body condition scoring (BCS) is an instrument-free method for visually inspecting and palpating lean and fat tissue. Animals are rated for lean and fat mass abundance, using an ordinal scale with species-specific descriptions as reference. However, modified BCS systems developed for other species are being used, because no BCS system has been fully validated for marmosets. The accuracy of BCS in identifying marmosets with poor body condition or obesity remains unknown. We assessed an adapted BCS for marmosets (n=68, 2-16 years). Objectives were to 1) determine whether BCS predicts body weight and body composition, and 2) evaluate the performance of BCS as a screening test for low body condition and obesity in marmosets, in comparison to QMR body composition analysis. BCS predicted body weight and body composition (F(15, 166)=7.51, Wilks {Lambda}=0.240, p<0.001), and was better at predicting low lean mass and obesity, than at predicting low adiposity. Marmosets with low BCS had higher odds of low lean mass (B=3.37, (95% CI, 0.95-5.78), OR=29.0, p=0.006). Marmosets with excessively high BCS had higher odds of obesity (B=2.72, (95% CI, 1.07-4.38), OR=15.23, p=0.001). The accuracy of BCS suggests it can serve as an instrument-free method to screen for low body condition (79.4%-91.2%) and obesity (77.9%) in marmosets. Research highlightsO_LIWe evaluated body condition scoring (BCS) as a screening tool for detecting low body condition and obesity in marmosets by comparing it to diagnoses based on quantitative magnetic resonance, the gold-standard method for body composition analysis. C_LIO_LIBCS was more accurate at detecting low lean mass and obesity than low adiposity, with marmosets having low BCS showing higher odds of low lean mass and those with excessively high BCS having higher odds of obesity. C_LIO_LIResults suggest that BCS can serve as an instrument-free method to screen for low body condition and obesity in marmosets, enabling early detection of health decline and guiding the need for further diagnostic testing and treatment. C_LI

physiology↗

bioRxiv2023

Detection of rare plasmid hosts using a targeted Hi-C approach

Despite the significant role plasmids play in microbial evolution, there is limited knowledge of their ecology, evolution, and transfer in microbial communities. Therefore, we developed and implemented a novel approach to identify rare plasmid hosts by combining Hi-C, a proximity ligation method, with enrichment for plasmid-specific DNA. We hereafter refer to this Hi-C enrichment approach as Hi-C+. Our experimental design mimicked scenarios in which the transfer of an antimicrobial resistance plasmid from a donor to a recipient in soil was increasingly rare. We established that Hi-C can link a plasmid to its host in soil when the relative abundance of that plasmid-host pair is as low as 0.001%. The Hi-C+ method further improved the detection limit of Hi-C 100-fold and allowed identification of plasmid hosts at the genus level. Therefore, Hi-C+ will facilitate the exploration of the ecological and evolutionary pathways that affect the spread of plasmids in natural environments. TeaserIn this study we demonstrate that a target-enriched Hi-C approach can identify rare hosts of a given plasmid in soil.

molecular biology↗

bioRxiv2022

Biofilms preserve transmissibility of a multi-drug resistance plasmid

Self-transmissible multidrug resistance (MDR) plasmids are a major health concern because they can spread antibiotic resistance to pathogens. Even though most pathogens form biofilms, little is known about how MDR plasmids persist and evolve in biofilms. We hypothesize that (i) biofilms act as refugia of MDR plasmids by retaining them in the absence of antibiotics longer than well-mixed planktonic populations, and that (ii) the evolutionary trajectories that account for the improvement of plasmid persistence over time differ between biofilms and planktonic populations. In this study, we evolved Acinetobacter baumannii with an MDR plasmid in biofilm and planktonic populations with and without antibiotic selection. In the absence of selection biofilm populations were better able to maintain the MDR plasmid than planktonic populations. In planktonic populations plasmid persistence improved rapidly but was accompanied by a loss of genes required for the horizonal transfer of plasmids. In contrast, in biofilms most plasmids retained their transfer genes, but on average plasmid persistence improved less over time. Our results showed that biofilms can act as refugia of MDR plasmids and favor the horizontal mode of plasmid transfer, which has important implications for the spread of MDR.

microbiology↗

bioRxiv2021

A Broadly Conserved Deoxycytidine Deaminase Protects Bacteria from Phage Infection

The El Tor biotype of Vibrio cholerae is responsible for perpetuating the longest cholera pandemic in recorded history (1961-current). The genomic islands VSP-1 and -2 are two understudied genetic features that distinguish El Tor from previous pandemics. To understand their utility, we calculated the co-occurrence of VSP genes across bacterial genomes. This analysis predicted the previously uncharacterized vc0175, herein renamed deoxycytidylate deaminase Vibrio (dcdV), is in a gene network with dncV, a cyclic GMP-AMP synthase involved in phage defense. DcdV consists of two domains, a P-loop kinase and a deoxycytidylate deaminase, that are required for the deamination of dCTP and dCMP, inhibiting phage predation by corrupting cellular nucleotide concentrations. Additionally, DcdV is post-translationally inhibited by a unique noncoding RNA encoded 5 of the dcdV locus. DcdV homologs are conserved in bacteria and eukaryotes and our results identify V. cholerae DcdV as the founding member of a previously undescribed bacterial phage defense system.

microbiology↗

bioRxiv2020

Trait Evolution in Microbial Communities

Understanding the evolutionary dynamics of microbial communities is a key step towards the goal of predicting and manipulating microbiomes to promote beneficial states. While interactions within microbiomes and between microbes and their environment collectively determine the community composition and population dynamics, we are often concerned with traits or functions of a microbiome that link more directly to host health. To study how traits of a microbiome are impacted by eco-evolutionary dynamics, we recast a classic resource-mediated population dynamic model into a population genetic framework which incorporates traits. The relative fitness of each group of microbes can be explicitly written in terms of population dynamic parameters, and corresponding evolutionary dynamics emerge. Using several example systems, we demonstrate how natural selection, mutation, and shifts in the environment work together to produce changes in traits over time.

microbiology↗

bioRxiv2020

Rhinovirus reduces the severity of subsequent respiratory viral infections, which is associated with dampened inflammatory responses

Coinfection by unrelated viruses in the respiratory tract is common and can result in changes in disease severity compared to infection by individual virus strains. We have previously shown that inoculation of mice with rhinovirus (RV) two days prior to inoculation with a lethal dose of influenza A virus (PR8), provides complete protection against mortality and reduces morbidity. In this study, we extended that finding to a second lethal respiratory virus, pneumonia virus of mice (PVM) and analyzed potential mechanisms whereby RV reduces lethal viral pneumonia caused by PR8 and PVM. RV prevented mortality and weight loss associated with PVM infection, suggesting that RV-mediated protection is more effective against PVM than PR8. Major changes in host gene expression upon PVM infection were delayed compared to PR8, which likely provides a larger time frame for RV-induced gene expression to alter the course of disease. Overall, RV induced earlier recruitment of inflammatory cells, while these populations were reduced at later times in RV-inoculated mice. Findings common to both virus pairs included upregulated expression of mucin-associated genes and dampening of inflammation-related genes in mice that were inoculated with RV prior to lethal virus infection. However, type I IFN signaling was required for RV-mediated protection against PR8, but not PVM. IFN signaling had minor effects on PR8 replication and contributed to controlling neutrophilic inflammation and subsequent hemorrhagic lung pathology in RV/PR8 infected mice. These findings, combined with differences in virus replication levels and disease severity, suggest that the suppression of inflammation in RV/PVM infected mice may be due to early, IFN-independent suppression of viral replication, while in RV/PR8 infected mice may be due to IFN-dependent modulation of immune responses. Thus, a mild upper respiratory viral infection can reduce the severity of a subsequent severe viral infection in the lungs through virus-dependent mechanisms. Author SummaryRespiratory viruses from diverse families co-circulate in human populations and are frequently detected within the same host. Though clinical studies suggest that infection by more than one unrelated respiratory virus may alter disease severity, animal models in which we can control the doses, timing, and strains of coinfecting viruses are critical to understand how coinfection affects disease severity. In this study, we compared gene expression and immune cell recruitment between two pairs of viruses (RV/PR8 and RV/PVM) inoculated sequentially in mice that both result in reduced severity compared to lethal infection by PR8 or PVM alone. Reduced disease severity was associated with suppression of inflammatory responses in the lungs. However, differences in disease kinetics and host and viral gene expression suggest that protection by coinfection with RV may be due to distinct molecular mechanisms. Indeed, we found that antiviral cytokine signaling was required for RV-mediated protection against lethal infection by PR8, but not PVM.

microbiology↗

bioRxiv2019

Microbial phenotypic heterogeneity in response to a metabolic toxin: continuous, dynamically shifting distribution of formaldehyde tolerance in Methylobacterium extorquens populations

While microbiologists often make the simplifying assumption that genotype determines phenotype in a given environment, it is becoming increasingly apparent that phenotypic heterogeneity (in which one genotype generates multiple phenotypes simultaneously even in a uniform environment) is common in many microbial populations. The importance of phenotypic heterogeneity has been demonstrated in a number of model systems involving binary phenotypic states (e.g., growth/non-growth); however, less is known about systems involving phenotype distributions that are continuous across an environmental gradient, and how those distributions change when the environment changes. Here, we describe a novel instance of phenotypic diversity in tolerance to a metabolic toxin within wild-type populations of Methylobacterium extorquens, a ubiquitous phyllosphere methylotroph capable of growing on the methanol periodically released from plant leaves. The first intermediate in methanol metabolism is formaldehyde, a potent cellular toxin that is lethal in high concentrations. We have found that at moderate concentrations, formaldehyde tolerance in M. extorquens is heterogeneous, with individual cells minimum tolerance levels ranging between 0 mM and 8 mM. This form of heterogeneity is continuous in terms of threshold (in the range of maximum tolerances possible), yet binary in outcome (at a given formaldehyde concentration, cells either grow normally or die, with no intermediate phenotype), and it is not associated with any detectable genetic mutations. Moreover, tolerance distributions within the population are dynamic, changing over time in response to growth conditions. We characterized this phenomenon using bulk liquid culture experiments, colony growth tracking, flow cytometry, time-lapse microscopy, and genome resequencing. Finally, we used mathematical modeling to better understand the processes by which cells change phenotype, and found evidence for both stochastic, bidirectional phenotype diversification and responsive, directed phenotype shifts, depending on the growth substrate and the presence of toxin.

microbiology↗

bioRxiv2018

Parameter Identifiability of the Generalized Lotka-Volterra Model for Microbiome Studies

BackgroundPopulation dynamic models can be used in conjunction with time series of species abundances to infer interactions. Understanding microbial interactions is a prerequisite for numerous goals in microbiome research; predicting how populations change over time, determining how manipulations of microbiomes affect dynamics, and designing synthetic microbiomes to perform tasks are just a few examples. As such, there is great interest in adapting population dynamic theory for microbial systems. Despite the appeal, numerous hurdles exist. One hurdle is that the data commonly obtained from DNA sequencing yield estimates of relative abundances, while population dynamic models such as the generalized Lotka-Volterra model track absolute abundances or densities. It is not clear whether relative abundance data alone can be used to infer parameters of population dynamic models such as the Lotka-Volterra model.\n\nResultsWe used structural identifiability analyses to determine the extent to which time series of relative abundances can be used to parameterize the generalized Lotka-Volterra model. We found that only with absolute abundance data to accompany relative abundance estimates from sequencing can all parameters be uniquely identified. However, relative abundance data alone do contain information on relative interaction strengths, which is sufficient for many studies where the goal is to estimate key interactions and their effects on dynamics. Our results also indicate that the relative interaction rates that can be estimated using relative abundance data provide ample information to estimate relative changes of absolute abundance over time. Using synthetic data for which we know the underlying structure, we found our results to be robust to modest amounts of both process and measurement error.\n\nConclusionsFitting the generalized Lotka-Volterra model to time-series sequencing data typically requires either assuming a constant population size or performing additional measurements to obtain absolute abundances. We have found that these assumptions are not strictly necessary because relative abundance data alone contain sufficient information to estimate relative rates of interaction, and thus to infer key drivers of microbial population dynamics.

microbiology↗

bioRxiv2016

Lung epithelial cells have virus-specific and shared gene expression responses to infection by diverse respiratory viruses

The severity and outcome of respiratory viral infections is partially determined by the cellular response mounted by infected lung epithelial cells. Disease prevention and treatment is dependent on our understanding of the shared and unique responses elicited by diverse viruses, yet few studies compare host responses to different viruses while controlling other experimental parameters. We compared changes in gene expression of murine lung epithelial cells infected individually by three respiratory viruses causing mild (rhinovirus, RV1B), moderate (coronavirus, MHV-1), and severe (influenza A virus, PR8) disease in mice. RV1B infection caused numerous gene expression changes, but the differential effect peaked at 12 hours post-infection. PR8 altered an intermediate number of genes whose expression continued to change through 24 hours. MHV-1 had comparatively few effects on host gene expression. The viruses elicited highly overlapping responses in antiviral genes, though MHV-1 induced a lower type I interferon response than the other two viruses. Signature genes were identified for each virus and included host defense genes for PR8, tissue remodeling genes for RV1B, and transcription factors for MHV-1. Our comparative approach identified universal and specific transcriptional signatures of virus infection that can be used to discover mechanisms of pathogenesis in the respiratory tract.

immunology↗