Statistical methods for detecting differentially abundant features in clinical metagenomic samples - PubMed (original) (raw)

Statistical methods for detecting differentially abundant features in clinical metagenomic samples

James Robert White et al. PLoS Comput Biol. 2009 Apr.

Abstract

Numerous studies are currently underway to characterize the microbial communities inhabiting our world. These studies aim to dramatically expand our understanding of the microbial biosphere and, more importantly, hope to reveal the secrets of the complex symbiotic relationship between us and our commensal bacterial microflora. An important prerequisite for such discoveries are computational tools that are able to rapidly and accurately compare large datasets generated from complex bacterial communities to identify features that distinguish them.We present a statistical method for comparing clinical metagenomic samples from two treatment populations on the basis of count data (e.g. as obtained through sequencing) to detect differentially abundant features. Our method, Metastats, employs the false discovery rate to improve specificity in high-complexity environments, and separately handles sparsely-sampled features using Fisher's exact test. Under a variety of simulations, we show that Metastats performs well compared to previously used methods, and significantly outperforms other methods for features with sparse counts. We demonstrate the utility of our method on several datasets including a 16S rRNA survey of obese and lean human gut microbiomes, COG functional profiles of infant and mature gut microbiomes, and bacterial and viral metabolic subsystem data inferred from random sequencing of 85 metagenomes. The application of our method to the obesity dataset reveals differences between obese and lean subjects not reported in the original study. For the COG and subsystem datasets, we provide the first statistically rigorous assessment of the differences between these populations. The methods described in this paper are the first to address clinical metagenomic datasets comprising samples from multiple subjects. Our methods are robust across datasets of varied complexity and sampling level. While designed for metagenomic applications, our software can also be applied to digital gene expression studies (e.g. SAGE). A web server implementation of our methods and freely available source code can be found at http://metastats.cbcb.umd.edu/.

PubMed Disclaimer

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Figure 1

Figure 1. Format of the feature abundance matrix.

Each row represents a specific taxon, while each column represents a subject or replicate. The frequency of the i th feature in the j th subject (c(i,j)) is recorded in the corresponding cell of the matrix. If there are g subjects in the first population, they are represented by the first g columns of the matrix, while the remaining columns represent subjects from the second population.

Figure 2

Figure 2. Detecting differential abundance for sparse features.

A 2×2 contingency table is used in Fisher's exact test for differential abundance between rare features. f11 is the number of observations of feature i in all individuals from treatment 1. f21 is the number of observations that are not feature i in all individuals from treatment 1. f12 and f22 are similarly defined for treatment 2.

Figure 3

Figure 3. Dispersion estimates (φ) for three metagenomic datasets used in this study.

These plots compare dispersion values between (A) obese and lean human gut taxonomic data, (B) infant and mature human gut COG assignments, and (C) microbial and viral subsystem annotations. We find a wide range of possible dispersions in this data and significant differences in dispersions between two populations.

Figure 4

Figure 4. ROC curves comparing statistical methods in a simulation study.

Sequences were selected from a beta-binomial distribution with variable dispersions and group mean proportions p1 and p2. For each set of parameters, we simulated 1000 trials, 500 of which are generated under the null hypothesis (p1 = p2), and the remainder are differentially abundant where a*p1 = p2. For example, p = 0.2 and a = 2 indicates features comprising 20% of the population that differ two-fold in abundance between two populations of interest. Parameter values for p1 and a are shown above each plot.

Figure 5

Figure 5. ROC curves comparing statistical methods in a simulation study for extreme sparse sampling.

Sequences were selected from a beta-binomial distribution with variable dispersions and group mean proportions p1 and p2. For each set of parameters, we simulated 1000 trials, 500 of which are generated under the null hypothesis (p1 = p2), and the remainder are differentially abundant where a*p1 = p2. For example, p = 0.2 and a = 2 indicates features comprising 20% of the population that differ two-fold in abundance between two populations of interest. Parameter values for p1 and a are shown above each plot.

Similar articles

Cited by

References

    1. Schloss PD, Handelsman J. Metagenomics for studying unculturable microorganisms: cutting the Gordian knot. Genome Biol. 2005;6:229. - PMC - PubMed
    1. Bik EM, Eckburg PB, Gill SR, Nelson KE, Purdom EA, et al. Molecular analysis of the bacterial microbiota in the human stomach. Proc Natl Acad Sci U S A. 2006;103:732–737. - PMC - PubMed
    1. Batzoglou S, Jaffe DB, Stanley K, Butler J, Gnerre S, et al. ARACHNE: a whole-genome shotgun assembler. Genome Res. 2002;12:177–189. - PMC - PubMed
    1. Palmer C, Bik EM, Digiulio DB, Relman DA, Brown PO. Development of the Human Infant Intestinal Microbiota. PLoS Biol. 2007;5:e177. doi:10.1371/journal.pbio.0050177. - PMC - PubMed
    1. Sogin ML, Morrison HG, Huber JA, Welch DM, Huse SM, et al. Microbial diversity in the deep sea and the underexplored “rare biosphere”. Proc Natl Acad Sci U S A. 2006;103:12115–12120. - PMC - PubMed

Publication types

MeSH terms

Substances

LinkOut - more resources