Tax4Fun: predicting functional profiles from metagenomic 16S rRNA data.

Aßhauer KP, Wemheuer B, Daniel R, Meinicke P. · 2015

Read Tax4Fun: predicting functional profiles from metagenomic 16S rRNA data. by Aßhauer KP, Wemheuer B, Daniel R, Meinicke P. (2015) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc4547618.txt.

Introduction

Kathrin P. Aßhauer, Bernd Wemheuer, Rolf Daniel, and Peter Meinicke present software that guesses what a microbe community can do, in Bioinformatics in 2015. Sequencing the 16S rRNA marker gene names the microbes in a sample cheaply, but says nothing direct about their abilities. Full metagenome sequencing answers that, at far higher cost.

Their package, Tax4Fun, predicts a community's functional profile from 16S data alone. Tested on paired datasets where both kinds of sequencing existed, its profiles came close to the shotgun results. The tool is an open-source R package that works with standard analysis outputs. For broad surveys with many samples, it offers function-level insight at marker-gene prices.

Catalog record

CREATORAßhauer KP, Wemheuer B, Daniel R, Meinicke P.
YEAR2015
COLLECTIONPAPERS
FORMATTXT
FILE SIZE13 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC4547618

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Tax4Fun: predicting functional profiles from metagenomic 16S rRNA data By Kathrin P. Aßhauer, Bernd Wemheuer, Rolf Daniel, Peter Meinicke Published in: Bioinformatics DOI: 10.1093/bioinformatics/btv287 ABSTRACT Motivation: The characterization of phylogenetic and functional diversity is a key element in the analysis of microbial communities. Amplicon-based sequencing of marker genes, such as 16S rRNA, is a powerful tool for assessing and comparing the structure of microbial communities at a high phylogenetic resolution. Because 16S rRNA sequencing is more cost-effective than whole metagenome shotgun sequencing, marker gene analysis is frequently used for broad studies that involve a large number of different samples. However, in comparison to shotgun sequencing approaches, insights into the functional capabilities of the community get lost when restricting the analysis to taxonomic assignment of 16S rRNA data. Results: Tax4Fun is a software package that predicts the functional capabilities of microbial communities based on 16S rRNA datasets. We evaluated Tax4Fun on a range of paired metagenome/16S rRNA datasets to assess its performance. Our results indicate that Tax4Fun provides a good approximation to functional profiles obtained from metagenomic shotgun sequencing approaches. Availability and implementation: Tax4Fun is an open-source R package and applicable to output as obtained from the SILVAngs web server or the application of QIIME with a SILVA database extension. Tax4Fun is freely available for download at http://tax4fun.gobics.de/. Contact: [email protected] Supplementary information: Supplementary data are available at Bioinformatics online. 1 INTRODUCTION Amplicon-based sequencing of marker genes is widely used for large-scale studies that involve many different sampling sites or time series. The common 16S rRNA gene-based analysis is a powerful tool for assessing the phylogenetic distribution of a metagenome but does not provide insights into the communities metabolic potential. Therefore, the prediction of the functional capabilities of a microbial community based on marker gene data would be highly beneficial. As a particular difficulty of such a predictive approach for most organisms in marker gene databases the genome and therefore the functional repertoire is not known. For instance, the SILVA SSU rRNA database (Quast et al., 2013) (SILVA 115 full release) contains 3 808 884 rRNA sequences whereas KEGG (Release 71.1) (Kanehisa et al., 2014) only comprises 2982 complete prokaryotic genomes. In a previous study (Aßhauer and Meinicke, 2013), we introduced a statistical method for predicting the metabolic profiles of a metagenome from its taxonomic composition using a linear combination of precomputed genomic reference profiles. A similar method has been proposed for inference of the community structure from remote sensing satellite image models (Larsen et al., 2015). In this approach, the average EC number counts of all annotated genomes from a given taxonomic group, e.g. at order level, are used as reference for the linear combination of the community structure at this level. Recently, the PICRUSt approach was proposed to predict KEGG Ortholog (KO) functional profiles of microbial communities using 16S rRNA gene sequences (Langille et al., 2013). PICRUSt infers unknown gene content by an extended ancestral state reconstruction algorithm. The algorithm uses a phylogenetic tree of 16S rRNA gene sequences to link operational taxonomic units (OTUs) with gene content. Thus, PICRUSt predictions depend on the topology of the tree and the distance to the next sequenced organism. Because a nearest neighbor within the tree topology always exists, PICRUSt links all OTUs, even if distances are large. This procedure can be problematic when analyzing microbial communities with a large proportion of so far not well-characterized phyla. Here, we present Tax4Fun, a novel tool for functional community profiling based on 16S rRNA data. In Tax4Fun the linking of 16S rRNA gene sequences with the functional annotation of sequenced prokaryotic genomes is realized with a nearest neighbor identification based on a minimum 16S rRNA sequence similarity. Tax4Fun can be applied to the output of 16S rRNA analysis pipelines that can perform a mapping of 16S rRNA gene reads to SILVA. The results of Tax4Fun indicate that the correlation of functional predictions with the metagenome profile is higher as compared to the PICRUSt tool. 2 IMPLEMENTATION Our method provides a prediction of functional profiles on the basis of SILVA-labeled OTU abundances. After preprocessing and clustering of the 16S rRNA sequencing reads the resulting OTUs have to be assigned to reference sequences in the SILVA database. The SILVA assignment counts are then transformed to functional profiles using Tax4Fun, which proceeds in three steps. First, the SILVA-based 16S rRNA profile is transformed to a taxonomic profile of the prokaryotic KEGG organisms. The linear transformation is realized by a precomputed association matrix (see Supplementary Material section 2.1.1). Then, the estimated abundances of KEGG organisms are normalized by the 16S rRNA copy number obtained from the NCBI genome annotations. Finally, the normalized taxonomic abundances are used to linearly combine the precomputed functional profiles of the KEGG organisms for the prediction of the functional profile of the microbial community. The organism-specific reference profiles are estimated with the same method as used for the Taxy-Pro reference profiles (Klingenberg et al., 2013). For a fast computation of the organism-specific and metagenomic functional KEGG Ortholog (KO) profiles, we utilized UProC (Meinicke, 2015) and PAUDA (Huson and Xie, 2014), respectively (see Supplementary Material section 2.1.2). 3 RESULTS We applied Tax4Fun and PICRUSt to a collection of paired metagenome/16S rRNA datasets that have also been used in the original PICRUSt study (Fierer et al., 2012; Harris et al., 2013; Human Microbiome Project Consortium, 2012; Kunin et al., 2008; Muegge et al., 2011) (see Supplementary Material section 1.1). Before applying Tax4Fun, the SILVA-based 16S rRNA profiles were computed using the QIIME tool (Caporaso et al., 2010) or the SILVAngs web server (Quast et al., 2013), respectively (see Supplementary Material section 2.2 and 2.3). For each paired dataset, the Spearman correlation of the whole metagenome and the 16S rRNA-predicted relative KO abundance profile was calculated. For the computation of the correlation, we excluded for each dataset all KO profile dimensions that did not contribute any non-zero count in the functional profiles. The resulting correlation coefficients are shown in Figure 1 for the UProC-based functional profiles. Using Tax4Fun, the median of the correlation coefficient varies between 0.8706 (soils) and 0.6427 (Guerrero Negro hypersaline microbial mat). Fig. 1.Spearman correlations between metagenomic and 16S-predicted functional profiles for comparison of Tax4Fun and PICRUSt on paired datasets from the human microbiome (HMP), mammalian guts, Guerrero Negro hypersaline microbial mat and soils In comparison with PICRUSt the correlation of Tax4Fun is significantly higher for all four datasets according to a nonparametric sign test (P-value < 0.001). Similar results are obtained using the PAUDA tool for estimation of the functional profiles (see Supplementary Material section 4.2). Further, we compared the coverage of the analysis pipelines in terms of the fraction of reads that were classified by QIIME/SILVAngs and the percentage of OTUs that were mapped to KEGG organisms using Tax4Fun. Especially for the soil samples, we observed rather low fractions of 16S rRNA sequences that were finally used to predict the functional profiles (SILVAngs: 0.02%, Tax4Fun: 4.78%; QIIME: 95.21%, Tax4Fun: 55.36%). Contrary, the coverage for the human microbiome and mammalian guts datasets is rather high for both QIIME/SILVAngs and Tax4Fun (SILVAngs: 95%, Tax4Fun 95%). In our study, the soil samples are probably the most complex communities under investigation. Further, our results revealed that members

Read the complete text