Population Structure Discovery in Meta-Analyzed Microbial Communities and Inflammatory Bowel Disease

Ma, Siyuan; Shungin, Dmitry; Mallick, Himel; Schirmer, Melanie; Nguyen, Long H.; Kolde, Raivo; Franzosa, Eric A.; Vlamakis, Hera; Xavier, Ramnik J.; Huttenhower, Curtis

doi:10.1101/2020.08.31.261214

Cited by 21 publications

(30 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is to be noted that MaAsLin 2’s capability extends well beyond association analysis. For instance, MaAsLin 2’s multi-analysis framework has been used in the context of meta-analysis 49 , and the extracted residuals and random effects from a MaAsLin 2 fit can be used for further downstream analysis (e.g. as has been done in the original HMP2 study for cross-measurement correlation analysis 36 ).…”

Section: Discussionmentioning

confidence: 99%

Multivariable Association Discovery in Population-scale Meta-omics Studies

Mallick

Rahnavard

McIver

et al. 2021

Preprint

Self Cite

296

204

View full text Add to dashboard Cite

It is challenging to associate features such as human health outcomes, diet, environmental conditions, or other metadata to microbial community measurements, due in part to their quantitative properties. Microbiome multi-omics are typically noisy, sparse (zero-inflated), high-dimensional, extremely non-normal, and often in the form of count or compositional measurements. Here we introduce an optimized combination of novel and established methodology to assess multivariable association of microbial community features with complex metadata in population-scale observational studies. Our approach, MaAsLin 2 (Microbiome Multivariable Associations with Linear Models), uses general linear models to accommodate a wide variety of modern epidemiological studies, including cross-sectional and longitudinal designs, as well as a variety of data types (e.g. counts and relative abundances) with or without covariates and repeated measurements. To construct this method, we conducted a large-scale evaluation of a broad range of scenarios under which straightforward identification of meta-omics associations can be challenging. These simulation studies reveal that MaAsLin 2’s linear model preserves statistical power in the presence of repeated measures and multiple covariates, while accounting for the nuances of meta-omics features and controlling false discovery. We also applied MaAsLin 2 to a microbial multi-omics dataset from the Integrative Human Microbiome (HMP2) project which, in addition to reproducing established results, revealed a unique, integrated landscape of inflammatory bowel disease (IBD) across multiple time points and omics profiles.

show abstract

Section: Discussionmentioning

confidence: 99%

Multivariable Association Discovery in Population-scale Meta-omics Studies

Mallick

Rahnavard

McIver

et al. 2021

Preprint

Self Cite

296

204

View full text Add to dashboard Cite

show abstract

“…These differences add non-biological variation to the samples, decreasing the ability to ascertain biological signals [69]. Various approaches have been proposed to remove technical artifacts from dataset collections, of which we selected two relevant to microbiome data [37, 38]. First, zero-centering methods aim to reduce batch effects by centering the mean of each feature within a batch to zero.…”

Section: Resultsmentioning

confidence: 99%

“…MMUPHin estimates parameters for the additive and multiplicative batch effects, using normal and inverse gamma distributions, respectively. The estimated parameters are then used to remove the batch effects from the dataset [38, 56]. For MMUPHin, the sample type (stool/biopsy) was used as a covariate for MMUPHin #1 and the sample type and disease label (UC/CD/Control) were covariates for MMUPHin #2.…”

Section: Methodsmentioning

confidence: 99%

“…For example, microbiome data’s compositional nature prevents the direct application of standard statistical methods as they may lead to erroneous results, and requires prior application of compositional normalization methods [33, 34]. In addition, methods have been developed to remove the technical “batch effects” commonly identified in collections of samples from different studies, such as naive zero-centering methods and the recently developed empirical Bayes’ method, Meta-analysis Methods with a Uniform Pipeline for Heterogeneity (MMUPHin) [35–38]. To date, the effect of various combinations of normalization and batch effect reduction techniques on ML model generalizability remains to be benchmarked.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Benchmark of data processing methods and machine learning models for gut microbiome-based diagnosis of inflammatory bowel disease

Kubinski¹,

Djamen-Kepaou²,

Zhanabaev³

et al. 2021

Preprint

View full text Add to dashboard Cite

BackgroundInflammatory bowel disease (IBD) patients wait months and undergo numerous invasive procedures between the initial appearance of symptoms and receiving a diagnosis. In order to reduce time until diagnosis and improve patient wellbeing, machine learning algorithms capable of diagnosing IBD from the gut microbiome’s composition are currently being explored. To date, these models have had limited clinical application due to decreased performance when applied to a new cohort of patient samples. Various methods have been developed to analyze microbiome data which may improve the generalizability of machine learning IBD diagnostic tests. With an abundance of methods, there is a need to benchmark the performance and generalizability of various machine learning pipelines (from data processing to training a machine learning model) for microbiome-based IBD diagnostic tools.ResultsWe collected fifteen 16S rRNA microbiome datasets (7707 samples) from North America to benchmark combinations of gut microbiome features, data normalization methods, batch effect reduction methods, and machine learning models. Pipeline generalizability to new cohorts of patients was evaluated with four binary classification metrics following leave-one dataset-out cross validation, where all samples from one study were left out of the training set and tested upon. We demonstrate that taxonomic features obtained from QIIME2 lead to better classification of samples from IBD patients than inferred functional features obtained from PICRUSt2. In addition, machine learning models that identify non-linear decision boundaries between labels are more generalizable than those that are linearly constrained. Prior to training a non-linear machine learning model on taxonomic features, it is important to apply a compositional normalization method and remove batch effects with the naive zero-centering method. Lastly, we illustrate the importance of generating a curated training dataset to ensure similar performance across patient demographics.ConclusionsThese findings will help improve the generalizability of machine learning models as we move towards non-invasive diagnostic and disease management tools for patients with IBD.

show abstract

“…All figures were drawn using ggplot2 from tidyverse package set [ 81 ]. MMUPHin R package was used for batch correction [ 82 ]. Shannon and Simpson alpha (within-sample) diversity indices were calculated using DivNet [ 83 ].…”

Section: Methodsmentioning

confidence: 99%

Comparative Evaluation of Sucrosomial Iron and Iron Oxide Nanoparticles as Oral Supplements in Iron Deficiency Anemia in Piglets

Mazgaj

Lipiński

Szudzik

et al. 2021

IJMS

View full text Add to dashboard Cite

Iron deficiency is the most common mammalian nutritional disorder. However, among mammalian species iron deficiency anemia (IDA), occurs regularly only in pigs. To cure IDA, piglets are routinely injected with high amounts of iron dextran (FeDex), which can lead to perturbations in iron homeostasis. Here, we evaluate the therapeutic efficacy of non-invasive supplementation with Sucrosomial iron (SI), a highly bioavailable iron supplement preventing IDA in humans and mice and various iron oxide nanoparticles (IONPs). Analysis of red blood cell indices and plasma iron parameters shows that not all iron preparations used in the study efficiently counteracted IDA comparable to FeDex-based supplementation. We found no signs of iron toxicity of any tested iron compounds, as evaluated based on the measurement of several toxicological markers that could indicate the occurrence of oxidative stress or inflammation. Neither SI nor IONPs increased hepcidin expression with alterations in ferroportin (FPN) protein level. Finally, the analysis of the piglet gut microbiota indicates the individual pattern of bacterial diversity across taxonomic levels, independent of the type of supplementation. In light of our results, SI but not IONPs used in the experiment emerges as a promising nutritional iron supplement, with a high potential to correct IDA in piglets.

show abstract

Population Structure Discovery in Meta-Analyzed Microbial Communities and Inflammatory Bowel Disease

Cited by 21 publications

References 61 publications

Multivariable Association Discovery in Population-scale Meta-omics Studies

Multivariable Association Discovery in Population-scale Meta-omics Studies

Benchmark of data processing methods and machine learning models for gut microbiome-based diagnosis of inflammatory bowel disease

Comparative Evaluation of Sucrosomial Iron and Iron Oxide Nanoparticles as Oral Supplements in Iron Deficiency Anemia in Piglets

Contact Info

Product

Resources

About