Tax4Fun2: prediction of habitat-specific functional profiles and functional redundancy based on 16S rRNA gene sequences.

Wemheuer F, Taylor JA, Daniel R, Johnston E, Meinicke P, Thomas T, Wemheuer B. · 2020

Read Tax4Fun2: prediction of habitat-specific functional profiles and functional redundancy based on 16S rRNA gene sequences. by Wemheuer F, Taylor JA, Daniel R, Johnston E, Meinicke P, Thomas T, Wemheuer B. (2020) 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-pmc8067651.txt.

Introduction

Wemheuer and co-authors present Tax4Fun2, a tool that predicts what a microbial community does from its 16S rRNA genes. Sequencing those genes names the microbes in a sample. Names alone cannot say what the community can do. Prediction tools fill the gap, but they lean on public genome databases. Those databases often miss the microbes of the ecosystem under study.

Tax4Fun2 improves on PICRUSt and the first Tax4Fun in accuracy tests. Predicted profiles matched metagenome profiles from the same samples closely. Users can add their own habitat-specific genomes to sharpen the fit. The tool also predicts functional gene redundancy. That measure hints at how well a community withstands disturbance. Redundancy scores matched those from simulated communities. The R package runs on modest hardware, with no cluster needed.

Catalog record

CREATORWemheuer F, Taylor JA, Daniel R, Johnston E, Meinicke P, Thomas T, Wemheuer B.
YEAR2020
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC8067651

Browse the full library · About the Tower of Babel

Full text

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

Tax4Fun2: prediction of habitat-specific functional profiles and functional redundancy based on 16S rRNA gene sequences By Franziska Wemheuer, Jessica A Taylor, Rolf Daniel, Emma Johnston, Peter Meinicke, Torsten Thomas, Bernd Wemheuer Published in: Environmental Microbiome DOI: 10.1186/s40793-020-00358-7 ABSTRACT Abstract Background Sequencing of 16S rRNA genes has become a powerful technique to study microbial communities and their responses towards changing environmental conditions in various ecosystems. Several tools have been developed for the prediction of functional profiles from 16S rRNA gene sequencing data, because numerous questions in ecosystem ecology require knowledge of community functions in addition to taxonomic composition. However, the accuracy of these tools relies on functional information derived from genomes available in public databases, which are often not representative of the microorganisms present in the studied ecosystem. In addition, there is also a lack of tools to predict functional gene redundancy in microbial communities. Results To address these challenges, we developed Tax4Fun2, an R package for the prediction of functional profiles and functional gene redundancies of prokaryotic communities from 16S rRNA gene sequences. We demonstrate that functional profiles predicted by Tax4Fun2 are highly correlated to functional profiles derived from metagenomes of the same samples. We further show that Tax4Fun2 has higher accuracies than PICRUSt and Tax4Fun. By incorporating user-defined, habitat-specific genomic information, the accuracy and robustness of predicted functional profiles is substantially enhanced. In addition, functional gene redundancies predicted with Tax4Fun2 are highly correlated to functional gene redundancies determined for simulated microbial communities. Conclusions Tax4Fun2 provides researchers with a unique tool to predict and investigate functional profiles of prokaryotic communities based on 16S rRNA gene sequencing data. It is easy-to-use, platform-independent and highly memory-efficient, thus enabling researchers without extensive bioinformatics knowledge or access to high-performance clusters to predict functional profiles. Another unique feature of Tax4Fun2 is that it allows researchers to calculate the redundancy of specific functions, which is a potentially important measure of how resilient a community will be to environmental perturbation. Tax4Fun2 is implemented in R and freely available at https://github.com/bwemheu/Tax4Fun2. Keywords: Metagenomics, Functional predictions, 16S rRNA gene, Bioinformatics, Microbiome, Multifunctional redundancy, Ecosystem functioning BACKGROUND Microorganisms are key players in ecosystem functioning [1, 2]. For example, host-associated microorganisms significantly contribute to the health of their host organism, e.g., by providing essential nutrients or by enhancing the host’s resilience to pathogens or pests [3–5]. High-throughput sequencing of 16S rRNA genes is a powerful and widely used tool to study the taxonomic or phylogenetic composition of microbial communities in a variety of marine [6–8], terrestrial [9, 10] and host-associated [11–13] environments. However, numerous questions in biogeochemistry and ecology require knowledge of microbial community function, rather than, or in addition to, the taxonomic or phylogenetic composition [14]. Investigating the composition of microbial communities and their functional capabilities is of particular importance when the effect of changing environmental conditions or anthropogenic perturbations on ecosystem services is being assessed [15]. Many ecosystems are threatened by environmental perturbations. A key question in microbial ecology is whether, and to what degree, microbial communities contain functionally redundant members that may provide stability to ecosystem processes in the face of these perturbations [8, 16–18]. It has been proposed that the evaluation of multiple functions provide a more comprehensive picture on the role of biodiversity for maintaining ecosystem functions [19]. However, the simultaneousness assessment of multiple functions is time-consuming [20]. Some researchers have started to address this question by developing metatranscriptome-based [20] or metagenome-based [21] approaches for quantifying the multifunctional redundancy of microbial communities [20, 21]. To the best of our knowledge, no tools are currently available to provide a standardized method for the simultaneous calculation of functional redundancy for multiple functions. In past years, several freely available tools including PICRUSt [22], Tax4Fun [23], Piphillin [24], Faprotax [25] and paprica [26] have been developed for the prediction of functional profiles inferred from 16S rRNA gene sequence data. Although these tools cannot replace the functional assessment via metagenomic shotgun sequencing, they have provided unique insights into functional capabilities of prokaryotic communities in diverse habitats, such as microbial mats [27], soil [28–31], marine seawater [25, 26, 32, 33], sediment [8, 34], rumen [35, 36] and the plant endosphere [37, 38]. The predictive power of these tools relies on functional information derived from genomes available in public databases. This information is used to predict functional profiles even if no close reference genomes are present in these databases. However, available genomes do not necessarily represent the functional diversity present in the ecosystem investigated. This problem has motivated the development of predictive tools specific for the rumen microbiome [35] or marine microorganisms [25]. Given the rapidly increasing number of available genomes, in particular through metagenome-assisted genome binning [39], and that many research groups have access to unpublished, habitat-specific genomic information, the incorporation of such data is likely to enhance the accuracy of functional inferences. Here, we introduce Tax4Fun2, a new and improved version of Tax4Fun [23]. Tax4Fun2 is platform-independent, user-friendly and highly memory-efficient. It can incorporate habitat-specific and user-defined data. Although Tax4Fun2 focuses on prokaryotic data, eukaryotic data can also be incorporated. We show that the incorporation of habitat-specific data improves the practical utility of Tax4Fun2 for microbiome datasets from a wide range of ecosystems. Comparative analysis further shows that Tax4Fun2 has a higher accuracy than PICRUSt [22] and Tax4Fun [23]. Another unique feature of Tax4Fun2 is that it enables researchers to calculate the functional redundancy of multiple functions, which is critical for the prediction of how likely a specific ecosystem function is to be lost during environmental perturbation. This information might be important in ecosystem biomonitoring and the prioritisation of environmental management actions. Implementation Tax4Fun2 is provided as an R [40] package with a current default reference dataset build from 12,377 archaeal and bacterial genomes available through the NCBI RefSeq database. The current version is 1.1.5. Tax4Fun2 is platform-independent and enables researchers without extensive bioinformatics knowledge to predict functional profiles in an efficient and user-friendly manner. In the following sections, we provide further details in the generation of the reference data and the technical implementation of Tax4Fun2. Tax4Fun2 reference data Tax4Fun2 is supplied with two reference datasets (Ref99NR and Ref100NR) referring to the similarity threshold used during clustering of the 16S rRNA gene sequences. Each dataset consists of an association matrix with 16S rRNA gene reference sequences associated with functional reference profiles (number of entries in the full association matrix: 4584 and 18,479 for Ref99NR and Ref100NR, respectively). Due to its smaller size, the Ref99NR database is less hardware-demanding and predictions are generated faster compared to the Ref100NR. In order to obtain the default reference data, we included the function buildReferenceData, which will automatically download and build the reference data. The reference datasets were generated as follows: we downloaded all complete genomes and all genomes with the status ‘chromosome’ from NCBI RefSeq (assessed on 18 August 2018), resulting in 275 archaeal and 12,102 bacterial genomes. Barrnap version 0.9 (https://github.com/tseemann/barrnap) was used to identify and extract all 16S rRNA gene sequences. All rRNA gene sequences were subsequently concatenated into a single file, sorted by decreasing length and clustered using the UCLUST algorithm implemented in USEARCH version 10.240 [41] at 99 and 100% sequence similarity, respectively. The longest sequence of each cluster serves as

Read the complete text