Longitudinal Microbiome Studies Research Articles

EXPLANA: A user-friendly workflow for EXPLoratory ANAlysis and feature selection in cross-sectional and longitudinal microbiome studies.

Longitudinal microbiome studies (LMS) are increasingly common but have analytic challenges including non-independent data requiring mixed-effects models and large amounts of data that motivate exploratory analysis to identify factors related to outcome variables. Although change analysis (i.e. calculating deltas between values at different timepoints) can be powerful, how to best conduct these analyses is not always clear. For example, observational LMS measurements show natural fluctuations, so baseline might not be a reference of primary interest; whereas, for interventional LMS, baseline is a key reference point, often indicating the start of treatment. To address these challenges, we developed a feature selection workflow for cross-sectional and LMS that supports numerical and categorical data called EXPLANA (EXPLoratory ANAlysis). Machine-learning methods were combined with different types of change calculations and downstream interpretation methods to identify statistically meaningful variables and explain their relationship to outcomes. EXPLANA generates an interactive report that textually and graphically summarizes methods and results. EXPLANA had good performance on simulated data, with an average area under the curve (AUC) of 0.91 (range: 0.79-1.0, SD = 0.05), outperformed an existing tool (AUC: 0.95 vs. 0.56), and identified novel order-dependent categorical feature changes. EXPLANA is broadly applicable and simplifies analytics for identifying features related to outcomes of interest.

TENSOR REGRESSION FOR INCOMPLETE OBSERVATIONS WITH APPLICATION TO LONGITUDINAL STUDIES.

Multivariate longitudinal data are frequently encountered in practice such as in our motivating longitudinal microbiome study. It is of general interest to associate such high-dimensional, longitudinal measures with some univariate continuous outcome. However, incomplete observations are common in a regular study design, as not all samples are measured at every time point, giving rise to the so-called blockwise missing values. Such missing structure imposes significant challenges for association analysis and defies many existing methods that require complete samples. In this paper we propose to represent multivariate longitudinal data as a three-way tensor array (i.e., sample-by-feature-by-time) and exploit a parsimonious scalar-on-tensor regression model for association analysis. We develop a regularized covariance-based estimation procedure that effectively leverages all available observations without imputation. The method achieves variable selection and smooth estimation of time-varying effects. The application to the motivating microbiome study reveals interesting links between the preterm infant's gut microbiome dynamics and their neurodevelopment. Additional numerical studies on synthetic data and a longitudinal aging study further demonstrate the efficacy of the proposed method.

White-nose syndrome restructures bat skin microbiomes.

The skin microbiome is an essential line of host defense against pathogens, yet our understanding of microbial communities and how they change when hosts become infected is limited. We investigated skin microbial composition in three North American bat species (Myotis lucifugus, Eptesicus fuscus, and Perimyotis subflavus) that have been impacted by the infectious disease, white-nose syndrome, caused by an invasive fungal pathogen, Pseudogymnoascus destructans. We compared bacterial and fungal composition from 154 skin swab samples and 70 environmental samples using a targeted 16S rRNA and internal transcribed spacer amplicon approach. We found that for M. lucifugus, a species that experiences high mortality from white-nose syndrome, bacterial microbiome diversity was dramatically lower when P. destructans was present. Key bacterial families-including those potentially involved in pathogen defense-significantly differed in abundance in bats infected with P. destructans compared to uninfected bats. However, skin bacterial diversity was not lower in E. fuscus or P. subflavus when P. destructans was present despite populations of the latter species declining sharply from white-nose syndrome. The fungal species present on bats substantially overlapped with the fungal taxa present in the environment at the site where the bat was sampled, but fungal community composition was unaffected by the presence of P. destructans for any of the three bat species. This species-specific alteration in bat skin bacterial microbiomes after pathogen invasion may suggest a mechanism for the severity of white-nose syndrome in M. lucifugus but not for other bat species impacted by the disease.IMPORTANCEInherent complexities in the composition of microbiomes can often preclude investigations of microbe-associated diseases. Instead of single organisms being associated with disease, community characteristics may be more relevant. Longitudinal microbiome studies of the same individual bats as pathogens arrive and infect a population are the ideal experiment but remain logistically challenging; therefore, investigations like our approach that are able to correlate invasive pathogens to alterations within a microbiome may be the next best alternative. The results of this study potentially suggest that microbiome-host interactions may determine the likelihood of infection. However, the contrasting relationship between Pd and the bacterial microbiomes of Myotis lucifugus and Perimyotis subflavus indicate that we are just beginning to understand how the bat microbiome interacts with a fungal invader such as Pd.

EMBED: Essential MicroBiomE Dynamics, a dimensionality reduction approach for longitudinal microbiome studies

Dimensionality reduction offers unique insights into high-dimensional microbiome dynamics by leveraging collective abundance fluctuations of multiple bacteria driven by similar ecological perturbations. However, methods providing lower-dimensional representations of microbiome dynamics both at the community and individual taxa levels are not currently available. To that end, we present EMBED: Essential MicroBiomE Dynamics, a probabilistic nonlinear tensor factorization approach. Like normal mode analysis in structural biophysics, EMBED infers ecological normal modes (ECNs), which represent the unique orthogonal modes capturing the collective behavior of microbial communities. Using multiple real and synthetic datasets, we show that a very small number of ECNs can accurately approximate microbiome dynamics. Inferred ECNs reflect specific ecological behaviors, providing natural templates along which the dynamics of individual bacteria may be partitioned. Moreover, the multi-subject treatment in EMBED systematically identifies subject-specific and universal abundance dynamics that are not detected by traditional approaches. Collectively, these results highlight the utility of EMBED as a versatile dimensionality reduction tool for studies of microbiome dynamics.

Temporal Alignment of Longitudinal Microbiome Data.

A major challenge in working with longitudinal data when studying some temporal process is the fact that differences in pace and dynamics might overshadow similarities between processes. In the case of longitudinal microbiome data, this may hinder efforts to characterize common temporal trends across individuals or to harness temporal information to better understand the link between the microbiome and the host. One possible solution to this challenge lies in the field of “temporal alignment” – an approach for optimally aligning longitudinal samples obtained from processes that may vary in pace. In this work we investigate the use of alignment-based analysis in the microbiome domain, focusing on microbiome data from infants in their first years of life. Our analyses center around two main use-cases: First, using the overall alignment score as a measure of the similarity between microbiome developmental trajectories, and showing that this measure can capture biological differences between individuals. Second, using the specific matching obtained between pairs of samples in the alignment to highlight changes in pace and temporal dynamics, showing that it can be utilized to predict the age of infants based on their microbiome and to uncover developmental delays. Combined, our findings serve as a proof-of-concept for the use of temporal alignment as an important and beneficial tool in future longitudinal microbiome studies.

ARZIMM: A Novel Analytic Platform for the Inference of Microbial Interactions and Community Stability from Longitudinal Microbiome Study.

Dynamic changes of microbiome communities may play important roles in human health and diseases. The recent rise in longitudinal microbiome studies calls for statistical methods that can model the temporal dynamic patterns and simultaneously quantify the microbial interactions and community stability. Here, we propose a novel autoregressive zero-inflated mixed-effects model (ARZIMM) to capture the sparse microbial interactions and estimate the community stability. ARZIMM employs a zero-inflated Poisson autoregressive model to model the excessive zero abundances and the non-zero abundances separately, a random effect to investigate the underlining dynamic pattern shared within the group, and a Lasso-type penalty to capture and estimate the sparse microbial interactions. Based on the estimated microbial interaction matrix, we further derive the estimate of community stability, and identify the core dynamic patterns through network inference. Through extensive simulation studies and real data analyses we evaluate ARZIMM in comparison with the other methods.

Identifying temporal and spatial patterns of variation from multimodal data using MEFISTO

Factor analysis is a widely used method for dimensionality reduction in genome biology, with applications from personalized health to single-cell biology. Existing factor analysis models assume independence of the observed samples, an assumption that fails in spatio-temporal profiling studies. Here we present MEFISTO, a flexible and versatile toolbox for modeling high-dimensional data when spatial or temporal dependencies between the samples are known. MEFISTO maintains the established benefits of factor analysis for multimodal data, but enables the performance of spatio-temporally informed dimensionality reduction, interpolation, and separation of smooth from non-smooth patterns of variation. Moreover, MEFISTO can integrate multiple related datasets by simultaneously identifying and aligning the underlying patterns of variation in a data-driven manner. To illustrate MEFISTO, we apply the model to different datasets with spatial or temporal resolution, including an evolutionary atlas of organ development, a longitudinal microbiome study, a single-cell multi-omics atlas of mouse gastrulation and spatially resolved transcriptomics.

482. SARS-CoV-2 Prevalence in Feces of Very Young Children, A Longitudinal Study

BackgroundUnderstanding the disease burden of SARS- CoV-2 in young children has been challenging as the majority are asymptomatic or experience mild symptoms and were rarely tested. SARS-CoV-2 is traditionally detected through respiratory secretions but has also been reported in feces where shedding may continue for weeks after respiratory samples show resolution. We examined the prevalence of SARS-CoV-2 in already collected fecal samples from young children through the pandemic as well as associated demographic factors.MethodsAs part of an ongoing longitudinal microbiome study in Northern Virginia, serial stools samples were collected from infants before and throughout the Covid-19 pandemic. Reverse transcription quantitative-PCR detecting SARS-CoV-2 nucleocapsid gene in the N1 and N2 regions was performed. Penalized logistic regression models were developed to evaluate the association between fecal positivity and potential risk factors. ResultsThe overall prevalence of SARS-CoV-2 in infant feces was 1.69 % (13 samples) with a prevalence at delivery, 2, 6, 12 and 24 months of 0, 0, 2.56, 1.96, and 0.85 % respectively. Fecal positivity was first detected 31 days before the reported first case of Covid-19 in Northern Virginia; prevalence rates peaked in September at 4.5% (Figure 1). Only one infant who tested positive was symptomatic with COVID-19 21 days before his stool was collected. Of the 13 positive samples, 8 reported Hispanic ethnicity and 7 reported an essential worker (Table 1). Penalized logistic regression model showed association between Hispanic ethnicity and testing positive (OR 5.04 (95% CI 1.7 – 15.0)) that remained after controlling for the presences of an essential worker (OR 4.7 (95% CI 1.6 – 14.0)). ConclusionPrevalence of SARS- CoV-2 in infant stool correlated with the prevalence of COVID-19 during the pandemic, with higher rates in those of Hispanic ethnicity corelating with regional trends. Fecal positivity in asymptomatic infants even before quarantine restrictions supports the early but silent transmission of SARS-CoV-2. This study likely underestimates true prevalence rates as stool samples were stored without viral preservative. There are many socioeconomic factors that predispose to disease while ethnicity may be a mediating or confounding factorDisclosures All Authors: No reported disclosures

Clostridioides Difficile Carriage Rate Increases during Pediatric ALL Treatment

▪BackgroundClostridioides difficile infection (CDI) is frequent in pediatric patients with acute lymphoblastic leukemia (ALL). Studies have shown upwards of 20% positivity rate in CDI testing among pediatric oncology patients, and up to several percent in pediatric ALL patients in the first 180 days of diagnosis. Antibiotic usage has been variably linked to CDI positivity in these populations. As CDI testing is usually done in symptomatic patients, the question of C. difficile carriage versus CDI has not been addressed. We and others have shown that microbiome is altered in pediatric ALL patients and survivors. We conducted a longitudinal stool microbiome study in pediatric ALL patients and tested the hypothesis that alteration of the microbiome during ALL treatment promotes C. difficile carriage.MethodsChildren with ALL were prospectively recruited on a rolling basis and stool samples were collected at diagnosis (Dx) and at the end of induction (EOI), consolidation (EOC), interim maintenance I (IMI), delayed intensification (DI), interim maintenance II (IMII), as well as approximately 3 months and 6 months into maintenance (M3, M6). Stool samples from healthy siblings were used as controls. TaqMan-based quantitative-PCR (qPCR) was performed on DNA extracted from stool samples to detect C. diff 16S rRNA, tcdA, (Toxin A) and tcdB (Toxin B) genes . Samples positive or either tcdA or tcdB, or both, were designated positive for toxigenic C. difficile. 16S rRNA hypervariable region V4 was sequenced and analyzed for microbiome diversity and relative abundance of microbiota.Results32 ALL patients age 3 months-19 years were included. The diagnoses were 12 standard risk and 14 high risk pre-B ALL, 5 T-ALL, and 1 relapsed pre-B ALL. Stool samples were collected from 18 healthy siblings. The numbers of samples tested at each treatment phase were: 29 Dx, 24 EOI, 23 EOC, 25 IMI, 21 DI, 6 IMII, 14 M3, 7 M6. No patient had symptoms suggestive of CDI, and no patient was clinically tested or treated for CDI.Total number of stool samples tested was 149, of which 43 (29%) were positive for toxigenic C. difficile (Figure 1). At diagnosis, 2/29 (7%) patients were positive, compared to 2/18 (11%) in healthy siblings. At EOI, positivity rate increased to 17%, then up to 40% - 52% at EOC, IMI, and DI. C. difficile positivity were lower around 30% at M3 and M6, although few patients reached maintenance to contribute samples for analysis. Twenty-five patients (78%) were positive at some phase. Longitudinal analysis of individual patients showed that C. difficile positivity was intermittent through treatment phases; only 3 patients remained persistently positive. Seven patients (22%) were never positive.Multivariate analysis showed that EOC, IMI, and DI treatment phases were significant risk factors for C. difficile carriage. Neither the number of antibiotics nor the number of antibiotic courses administered was significant. Leukemia risk stratification (high risk versus standard risk) also did not correlate with C. difficile positivity.Microbiome analysis showed statistically significant differences in relative abundance of certain taxa between C. diff positive and negative samples at the class, order, and family levels (Figure 2). Examples include depletion of the class Verrucomicrobiae, which contains protective Akkermansia, and depletion of the common taxa Bifidobacteriaceae and Ruminococcaceae.ConclusionLongitudinal PCR testing of toxigenic C. difficile in pediatric ALL patients demonstrated increased C. difficile prevalence further into treatment phases. C. difficile carriage correlated significantly with depletion of several bacterial taxa, as microbiome diversity decreased overall with successive treatment phases. Our data lend support to the hypothesis that altered microbiome in ALL treatment allows permissibility for C. difficile carriage. In addition, no C. difficile positive patient had symptoms of CDI, therefore, caution must be taken in clinical testing, as there is a high asymptomatic carriage rate. Further longitudinal testing during maintenance and off-therapy is needed to see if C. difficile carriage rate returns to baseline and correlates with recovery of gut microbiome. [Display omitted] DisclosuresNo relevant conflicts of interest to declare.

"Longitudinal Fecal Microbiome Study of Total Body Irradiated Mice Treated With Radiation Mitigators Identifies Bacterial Associations With Survival".

Total body irradiation (TBI) has been demonstrated to alter the intestinal microbiome, but the effects of successful small molecule ionizing radiation mitigators on the intestinal microbiome are not well-known. Our survival experiments examined the effects of anti-cell death radiation mitigators on and in conjunction with the host’s microbiota. Mice received 9.25 Gy TBI and then were administered radiation mitigators 24 hours later. Passed stool were collected pre-irradiation, then on days 1, 3, 5, 7, 10, 14, 21, and 30 post-irradiation for 16S rRNA gene (V4 region) sequencing. The Cox proportional hazards (CPH) model was fit with taxonomic composition (time varying covariates) and treatment as predictors. In the first experiment, mice were administered drugs for “granulocyte stimulation and anti-apoptosis” in four protocol combinations: JP4-039 (anti-apoptosis), granulocyte colony-stimulating factor (G-CSF, granulopoietic precursor cell stimulator), both mitigators, and control. Survival improved relative to control (30.0%) for G-CSF (80%, p-value = 0.025), G-CSF/JP4-039 (70%, p-value = 0.084), but not for JP4-039 (50.0%). In the second experiment, mice were administered mitigation drugs “inhibiting programmed cell death” pathways: JP4-039 (anti-apoptosis), necrostatin-1 (anti-necroptosis), and baicalein (anti-ferroptosis), in eight combinations. The survival of JP4-039/baicalein (60.0%, p-value = 0.010) and JP4-039/baicalein/necrostatin-1 (60.0%, p-value = 0.06) treatment combinations were significantly different from the control (26.7%). The JP4-039/necrostatin-1 (46.7%) and baicalein/necrostatin-1 (40.0%) and singlet treatment combinations (26.7%) were not significantly different from the control. Despite differences between the baseline microbiota compositions of the two experiments, consistent changes in composition after irradiation were found: Lactobacillus decreased post-irradiation, relative to baseline. By day 7, microbiota perturbations had incompletely reversed, and no drug-specific differences were identifiable. The CPH model identified Lactobacillus and members of Ruminococcaceae, including Ruminococcus, as protective and Akkermansia as deleterious. By day 30, the microbiota of surviving mice had not returned to baseline, but the differences between experiments suggest the resultant microbiota composition of the survivors are stochastic or batch specific in nature, rather than a requirement for survival. In conclusion, the study determined that key taxa identified in fecal samples, when applied towards the prediction of TBI survival, improves the survival model relative to treatment information alone.

Microbial trend analysis for common dynamic trend, group comparison, and classification in longitudinal microbiome study

BackgroundThe human microbiome is inherently dynamic and its dynamic nature plays a critical role in maintaining health and driving disease. With an increasing number of longitudinal microbiome studies, scientists are eager to learn the comprehensive characterization of microbial dynamics and their implications to the health and disease-related phenotypes. However, due to the challenging structure of longitudinal microbiome data, few analytic methods are available to characterize the microbial dynamics over time.ResultsWe propose a microbial trend analysis (MTA) framework for the high-dimensional and phylogenetically-based longitudinal microbiome data. In particular, MTA can perform three tasks: 1) capture the common microbial dynamic trends for a group of subjects at the community level and identify the dominant taxa; 2) examine whether or not the microbial overall dynamic trends are significantly different between groups; 3) classify an individual subject based on its longitudinal microbial profiling. Our extensive simulations demonstrate that the proposed MTA framework is robust and powerful in hypothesis testing, taxon identification, and subject classification. Our real data analyses further illustrate the utility of MTA through a longitudinal study in mice.ConclusionsThe proposed MTA framework is an attractive and effective tool in investigating dynamic microbial pattern from longitudinal microbiome studies.

Longitudinal characterization of multispecies microbial populations recovered from spaceflight potable water

While sequencing technologies have revolutionized our knowledge of microbial diversity, little is known about the dynamic emergent phenotypes that arise within the context of mixed-species populations, which are not fully predicted using sequencing technologies alone. The International Space Station (ISS) is an isolated, closed human habitat that can be harnessed for cross-sectional and longitudinal functional microbiome studies. Using NASA-archived microbial isolates collected from the ISS potable water system over several years, we profiled five phenotypes: antibiotic resistance, metabolism, hemolysis, and biofilm structure/composition of individual or multispecies communities, which represent characteristics that could negatively impact astronaut health and life-support systems. Data revealed a temporal dependence on interactive behaviors, suggesting possible microbial adaptation over time within the ecosystem. This study represents one of the most extensive phenotypic characterization of ISS potable water microbiota with implications for microbial risk assessments of water systems in built environments in space and on Earth.

The link between the gut microbiota and Parkinson’s Disease: A systematic mechanism review with focus on α-synuclein transport

IntroductionResearch has suggested a link between the gut microbiota and Parkinson’s Disease (PD), and an early involvement of gastrointestinal dysfunction has been reported in patients. A mechanism review was performed to investigate whether the neurodegenerative cascade begins in the gut; mediated by gut dysbiosis and retrograde transport of α-synuclein. This review provides a summary of microbiome composition associated with PD, and evaluates pathophysiological mechanisms from animal and in vitro models of PD. MethodA systematic literature search was performed in PubMed; 82 of 299 papers met the inclusion criteria. ResultsAll twenty-two human case-control studies demonstrated an altered gut microbiota in PD compared to healthy controls, with results suggesting a proinflammatory phenotype present in PD. A germ-free animal study has demonstrated that gut microbiota are required for microglia activation, α-synuclein pathology and motor deficits. Accumulation of phosphorylated α-synuclein has been observed in the enteric nervous system prior to the onset of motor symptoms in animal models of PD, and there is data to support retrograde transport of α-synuclein from the gut to the brain. Different animal models of PD have demonstrated neuroinflammation, microglial activation and loss of dopaminergic neurons in the brain. ConclusionEvidence from this review supports the hypothesis that pathology spreads from the gut to the brain. Future animal studies using oral LPS or microbiota transplants from human PD cases could provide further insight into the entire mechanism. Prospective longitudinal microbiome studies and novel modelling approaches could help to identify functional dysbiosis and early biomarkers for PD.

Joint modeling of zero-inflated longitudinal proportions and time-to-event data with application to a gut microbiome study.

Recent studies have suggested that the temporal dynamics of the human microbiome may have associations with human health and disease. An increasing number of longitudinal microbiome studies, which record time to disease onset, aim to identify candidate microbes as biomarkers for prognosis. Owing to the ultra-skewness and sparsity of microbiome proportion (relative abundance) data, directly applying traditional statistical methods may result in substantial power loss or spurious inferences. We propose a novel joint modeling framework [JointMM], which is comprised of two sub-models: a longitudinal sub-model called zero-inflated scaled-beta generalized linear mixed-effects regression to depict the temporal structure of microbial proportions among subjects; and a survival sub-model to characterize the occurrence of an event and its relationship with the longitudinal microbiome proportions. JointMM is specifically designed to handle the zero-inflated and highly skewed longitudinal microbial proportion data and examine whether the temporal pattern of microbial presence and/or the nonzero microbial proportions are associated with differences in the time to an event. The longitudinal sub-model of JointMM also provides the capacity to investigate how the (time-varying) covariates are related to the temporal microbial presence/absence patterns and/or the changing trend in nonzero proportions. Comprehensive simulations and real data analyses are used to assess the statistical efficiency and interpretability ofJointMM.

PhyLoSTM: a novel deep learning model on disease prediction from longitudinal microbiome data.

Research shows that human microbiome is highly dynamic on longitudinal timescales, changing dynamically with diet, or due to medical interventions. In this article, we propose a novel deep learning framework 'phyLoSTM', using a combination of Convolutional Neural Networks and Long Short Term Memory Networks (LSTM) for feature extraction and analysis of temporal dependency in longitudinal microbiome sequencing data along with host's environmental factors for disease prediction. Additional novelty in terms of handling variable timepoints in subjects through LSTMs, as well as, weight balancing between imbalanced cases and controls is proposed. We simulated 100 datasets across multiple time points for model testing. To demonstrate the model's effectiveness, we also implemented this novel method into two real longitudinal human microbiome studies: (i) DIABIMMUNE three country cohort with food allergy outcomes (Milk, Egg, Peanut and Overall) and (ii) DiGiulio study with preterm delivery as outcome. Extensive analysis and comparison of our approach yields encouraging performance with an AUC of 0.897 (increased by 5%) on simulated studies and AUCs of 0.762 (increased by 19%) and 0.713 (increased by 8%) on the two real longitudinal microbiome studies respectively, as compared to the next best performing method, Random Forest. The proposed methodology improves predictive accuracy on longitudinal human microbiome studies containing spatially correlated data, and evaluates the change of microbiome composition contributing to outcome prediction. https://github.com/divya031090/phyLoSTM. Supplementary data are available at Bioinformatics online.

Longitudinal Survey of Fecal Microbiota in Healthy Dogs Administered a Commercial Probiotic.

The aim of this longitudinal microbiome study was to investigate the effects of a commercially available veterinary synbiotic product (Blackmore's® Paw DigestiCare 60™) on the fecal microbiome of healthy dogs using 16S rRNA gene microbial profiling. Fifteen healthy, privately-owned dogs participated in a 2-week trial administration of the product. Fecal samples were collected at different time points, including baseline (prior to treatment), during administration and after discontinuation of product. Large intra- and inter-individual variation was observed throughout the study, but microbiome composition at higher phylogenetic levels, alpha and beta diversity were not significantly altered after 2 weeks of probiotic administration, suggesting an absence of probiotic impact on microbial diversity. Administration of the synbiotic preparation did, however, result in transient increases in probiotic species from Enterococacceae and Streptococacceae families as well as an increase in Fusobacteria; with the fecal microbiota partially reverting to its baseline state 3-weeks after cessation of probiotic administration.

Challenges, Strategies, and Perspectives for Reference-Independent Longitudinal Multi-Omic Microbiome Studies.

In recent years, multi-omic studies have enabled resolving community structure and interrogating community function of microbial communities. Simultaneous generation of metagenomic, metatranscriptomic, metaproteomic, and (meta) metabolomic data is more feasible than ever before, thus enabling in-depth assessment of community structure, function, and phenotype, thus resulting in a multitude of multi-omic microbiome datasets and the development of innovative methods to integrate and interrogate those multi-omic datasets. Specifically, the application of reference-independent approaches provides opportunities in identifying novel organisms and functions. At present, most of these large-scale multi-omic datasets stem from spatial sampling (e.g., water/soil microbiomes at several depths, microbiomes in/on different parts of the human anatomy) or case-control studies (e.g., cohorts of human microbiomes). We believe that longitudinal multi-omic microbiome datasets are the logical next step in microbiome studies due to their characteristic advantages in providing a better understanding of community dynamics, including: observation of trends, inference of causality, and ultimately, prediction of community behavior. Furthermore, the acquisition of complementary host-derived omics, environmental measurements, and suitable metadata will further enhance the aforementioned advantages of longitudinal data, which will serve as the basis to resolve drivers of community structure and function to understand the biotic and abiotic factors governing communities and specific populations. Carefully setup future experiments hold great potential to further unveil ecological mechanisms to evolution, microbe-microbe interactions, or microbe-host interactions. In this article, we discuss the challenges, emerging strategies, and best-practices applicable to longitudinal microbiome studies ranging from sampling, biomolecular extraction, systematic multi-omic measurements, reference-independent data integration, modeling, and validation.

Dietary Patterns and Growth From 12 to 24 Months of Age in African American Infants

Abstract Objectives To identify dietary patterns in 12-month-old African American infants and investigate their association with change in infant BMI z-score (BMIz) from 12 to 24 months. Methods The prospective longitudinal Infant Growth and Microbiome Study (IGram) enrolled African American women in their 3rd trimester of pregnancy who had a pre-pregnancy BMI >30 or <25 and followed them and their infants from birth to 24 months of age. At 12 months, mothers reported infant intake of 32 food items in the past 7 days using the CDC Infant Feeding Practices Study II questionnaire. With these data we employed principal component analysis to derive dietary patterns for infants aged 12 months. We used multiple regression to test associations of dietary pattern scores with change in infant BMIz between 12 and 24 months. Results IGram enrolled 368 infants at birth; 320 and 283 infants completed 12- and 24-month visits, respectively. The prevalence of BMIz > 2SD was 7.8% at 12 months and 6% at 24 months. We derived three dietary patterns which together explained 36% of the variation in the dataset. Based on the highest factor loadings for each pattern, they were named “Transitional/table food,” “Formula/baby cereal,” and “Fruit and vegetable/breast milk.” The former pattern included pancakes, pasta, rice, french fries, meat, eggs, and snack and sweet foods. Regression results suggested that, accounting for maternal pre-pregnancy obesity status, the “Fruit and vegetable/breast milk” pattern explained a small portion of the variance in BMIz increase from 12 to 24 months of age (β = 0.07, 95% CI = 0.01, 0.13, P = 0.03). Conclusions Preliminary findings of three distinct dietary patterns in African American infants at 12 months of age are an important first step in characterizing infant feeding patterns across the rich, longitudinal IGram dataset. A positive association of the “Fruit and vegetable/breast milk” pattern with change in infant BMIz from 12 to 24 months was unexpected and requires further investigation. Funding Sources An unrestricted donation from the American Beverage Foundation for a Healthy America to support the Children's Hospital of Philadelphia's Healthy Weight Program; the Research Institute of the Children's Hospital of Philadelphia; the National Center for Advancing Translational Sciences; the NIH NIDDK.

Dynamic Bayesian Networks for Integrating Multi-omics Time Series Microbiome Data.

ABSTRACTA key challenge in the analysis of longitudinal microbiome data is the inference of temporal interactions between microbial taxa, their genes, the metabolites that they consume and produce, and host genes. To address these challenges, we developed a computational pipeline, a pipeline for the analysis of longitudinal multi-omics data (PALM), that first aligns multi-omics data and then uses dynamic Bayesian networks (DBNs) to reconstruct a unified model. Our approach overcomes differences in sampling and progression rates, utilizes a biologically inspired multi-omic framework, reduces the large number of entities and parameters in the DBNs, and validates the learned network. Applying PALM to data collected from inflammatory bowel disease patients, we show that it accurately identifies known and novel interactions. Targeted experimental validations further support a number of the predicted novel metabolite-taxon interactions.IMPORTANCE While a number of large consortia collect and profile several different types of microbiome and genomic time series data, very few methods exist for joint modeling of multi-omics data sets. We developed a new computational pipeline, PALM, which uses dynamic Bayesian networks (DBNs) and is designed to integrate multi-omics data from longitudinal microbiome studies. When used to integrate sequence, expression, and metabolomics data from microbiome samples along with host expression data, the resulting models identify interactions between taxa, their genes, and the metabolites that they produce and consume, as well as their impact on host expression. We tested the models both by using them to predict future changes in microbiome levels and by comparing the learned interactions to known interactions in the literature. Finally, we performed experimental validations for a few of the predicted interactions to demonstrate the ability of the method to identify novel relationships and their impact.

Modeling Longitudinal Microbiome Compositional Data: A Two-Part Linear Mixed Model with Shared Random Effects

Longitudinal microbiome studies have been widely used to unveil the dynamics in the complex host-microbial ecosystems. Modeling the longitudinal microbiome compositional data, which is semi-continuous in nature, is challenging in several aspects: the overabundance of zeros, the heavy skewness of non-zero values that are bounded in (0, 1), and the dependence between the binary and non-zero parts. To deal with these challenges, we first extended the work of Chen and Li [1] and proposed a two-part zero-inflated Beta regression model with shared random effects (ZIBR-SRE), which characterize the dependence between the binary and the continuous parts. Besides, the microbiome compositional data have unit-sum constraint, indicating the existence of negative correlations among taxa. As ZIBR-SRE models each taxon separately, it does not satisfy the sum-to-one constraint. We then proposed a two-part linear mixed model (TPLMM) with shared random effects to formulate the log-transformed standardized relative abundances rather than the original ones. Such transformation is called “additive logistic transformation”, initially developed for cross-sectional compositional data. We extended it to analyze the longitudinal microbiome compositions and showed that the unit-sum constraint can be automatically satisfied under the TPLMM framework. Model performances of TPLMM and ZIBR-SRE were compared with existing methods in simulation studies. Under settings adopted from real data, TPLMM had the best performance and is recommended for practical use. An oral microbiome application further showed that TPLMM and ZIBR-SRE estimated a strong correlation structure in the binary and the continuous parts, suggesting models without accounting for this dependence would lead to biased inferences.