Microbiome differential abundance methods produce different results across 38 datasets

Nearing, Jacob T.; Douglas, Gavin M.; Hayes, Molly G.; MacDonald, Jocelyn; Desai, Dhwani; McCormick, Nicole E.; Jones, C. M.; Wright, Robyn J.; Dhanani, Akhilesh S.; Comeau, A.; Langille, Morgan G. I.

doi:10.1038/s41467-022-28034-z

Cited by 416 publications

(426 citation statements)

References 84 publications

Supporting

Mentioning

339

Contrasting

Order By: Relevance

“…That said, modeling VoE is certainly not the only way to identify an association worth prioritizing. Furthermore, an ostensibly robust association viewed only through the lens of VoE still could be a false positive or dependent on, for example, data processing pipeline choice (e.g., the decision to average repeated measures data versus selecting one sample per individual) [ 37 ]. A more comprehensive framework could rely on a number of heuristics, for example, putting the greatest emphasis on associations that have the best model fit, are reported across multiple large cohorts, and/or have undergone sensitivity analysis via VoE.…”

Section: Discussionmentioning

confidence: 99%

Systematically assessing microbiome–disease associations identifies drivers of inconsistency in metagenomic research

et al. 2022

View full text Add to dashboard Cite

Evaluating the relationship between the human gut microbiome and disease requires computing reliable statistical associations. Here, using millions of different association modeling strategies, we evaluated the consistency—or robustness—of microbiome-based disease indicators for 6 prevalent and well-studied phenotypes (across 15 public cohorts and 2,343 individuals). We were able to discriminate between analytically robust versus nonrobust results. In many cases, different models yielded contradictory associations for the same taxon–disease pairing, some showing positive correlations and others negative. When querying a subset of 581 microbe–disease associations that have been previously reported in the literature, 1 out of 3 taxa demonstrated substantial inconsistency in association sign. Notably, >90% of published findings for type 1 diabetes (T1D) and type 2 diabetes (T2D) were particularly nonrobust in this regard. We additionally quantified how potential confounders—sequencing depth, glucose levels, cholesterol, and body mass index, for example—influenced associations, analyzing how these variables affect the ostensible correlation between Faecalibacterium prausnitzii abundance and a healthy gut. Overall, we propose our approach as a method to maximize confidence when prioritizing findings that emerge from microbiome association studies.

show abstract

Section: Discussionmentioning

confidence: 99%

Systematically assessing microbiome–disease associations identifies drivers of inconsistency in metagenomic research

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Investigations into the ability of statistical models to identify biomarkers has shown that different methods may not consistently detect the same biomarkers, likely due to differing assumptions about the underlying distribution of each ASV’s counts [ 29 , 30 ]. As a result, it is often difficult to choose a particular statistical model.…”

Section: Introductionmentioning

confidence: 99%

LANDMark: an ensemble approach to the supervised selection of biomarkers in high-throughput sequencing data

et al. 2022

View full text Add to dashboard Cite

Background Identification of biomarkers, which are measurable characteristics of biological datasets, can be challenging. Although amplicon sequence variants (ASVs) can be considered potential biomarkers, identifying important ASVs in high-throughput sequencing datasets is challenging. Noise, algorithmic failures to account for specific distributional properties, and feature interactions can complicate the discovery of ASV biomarkers. In addition, these issues can impact the replicability of various models and elevate false-discovery rates. Contemporary machine learning approaches can be leveraged to address these issues. Ensembles of decision trees are particularly effective at classifying the types of data commonly generated in high-throughput sequencing (HTS) studies due to their robustness when the number of features in the training data is orders of magnitude larger than the number of samples. In addition, when combined with appropriate model introspection algorithms, machine learning algorithms can also be used to discover and select potential biomarkers. However, the construction of these models could introduce various biases which potentially obfuscate feature discovery. Results We developed a decision tree ensemble, LANDMark, which uses oblique and non-linear cuts at each node. In synthetic and toy tests LANDMark consistently ranked as the best classifier and often outperformed the Random Forest classifier. When trained on the full metabarcoding dataset obtained from Canada’s Wood Buffalo National Park, LANDMark was able to create highly predictive models and achieved an overall balanced accuracy score of 0.96 ± 0.06. The use of recursive feature elimination did not impact LANDMark’s generalization performance and, when trained on data from the BE amplicon, it was able to outperform the Linear Support Vector Machine, Logistic Regression models, and Stochastic Gradient Descent models (p ≤ 0.05). Finally, LANDMark distinguishes itself due to its ability to learn smoother non-linear decision boundaries. Conclusions Our work introduces LANDMark, a meta-classifier which blends the characteristics of several machine learning models into a decision tree and ensemble learning framework. To our knowledge, this is the first study to apply this type of ensemble approach to amplicon sequencing data and we have shown that analyzing these datasets using LANDMark can produce highly predictive and consistent models.

show abstract

“…Nonetheless, the current POMS framework represents a more robust methodology compared with existing statistical tests applied in the microbiome field. Given the wide range of results across microbiome differential abundance approaches 31 , clearly more reliable approaches are needed, even if they come at the cost of lower sensitivity.…”

Section: Discussionmentioning

confidence: 99%

Integrating phylogenetic and functional data in microbiome studies

Douglas

Hayes

Langille

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Microbiome functional data are frequently analyzed to identify associations between microbial gene families and sample groups of interest. This is most often performed with approaches focused on the metagenome-wide relative abundance of microbial functions. Although such approaches can provide valuable insights, it is challenging to distinguish between different possible explanations for variation in community-wide functional profiles by considering functions alone. To help address this problem, we have developed a novel, phylogeny-aware framework to expand taxonomic balance tree-based approaches to identify enriched functions more robustly. The key focus of our approach, termed POMS, is on identifying functions that are consistently enriched in sample groups across independent taxonomic lineages. Based on simulated data we demonstrate that POMS can more accurately identify gene families that confer a selective advantage compared with commonly used differential abundance approaches. We also show that POMS can identify enriched functions in real-world metagenomics datasets that are potential targets of strong selection on multiple members of the microbiome. While this framework may not be able to identify all potential functional enrichments, the enrichments it does identify are more interpretable and conservative compared with those identified by existing differential abundance approaches. More generally, POMS is a novel approach for exploring microbiome functional data, which could be used to complement standard analyses. POMS is freely available as an R package at: https://github.com/gavinmdouglas/POMS.

show abstract

Microbiome differential abundance methods produce different results across 38 datasets

Cited by 416 publications

References 84 publications

Systematically assessing microbiome–disease associations identifies drivers of inconsistency in metagenomic research

Systematically assessing microbiome–disease associations identifies drivers of inconsistency in metagenomic research

LANDMark: an ensemble approach to the supervised selection of biomarkers in high-throughput sequencing data

Integrating phylogenetic and functional data in microbiome studies

Contact Info

Product

Resources

About