Biomartr: genomic data retrieval with R

Abstract

Motivation

Retrieval and reproducible functional annotation of genomic data are crucial in biology. However, the current poor usability and transparency of retrieval methods hinders reproducibility. Here we present an open source R package, biomartr, which provides a comprehensive easy-to-use framework for automating data retrieval and functional annotation for meta-genomic approaches. The functions of biomartr achieve a high degree of clarity, transparency and reproducibility of analyses.

Results

The biomartr package implements straightforward functions for bulk retrieval of all genomic data or data for selected genomes, proteomes, coding sequences and annotation files present in databases hosted by the National Center for Biotechnology Information (NCBI) and European Bioinformatics Institute (EMBL-EBI). In addition, biomartr communicates with the BioMart database for functional annotation of retrieved sequences. Comprehensive documentation of biomartr functions and five tutorial vignettes provide step-by-step instructions on how to use the package in a reproducible manner.

Availability and Implementation

The open source biomartr package is available at https://github.com/HajkD/biomartr and https://cran.r-project.org/web/packages/biomartr/index.html.

Supplementary information

Supplementary data are available at Bioinformatics online.

1 Introduction

Modern genome studies are no longer limited to single genome analyses or pairwise genomic comparisons but increasingly involve meta-genomic approaches. For this purpose, the NCBI and EMBL-EBI organize and maintain specialized sequence databases that fulfil various scientific requirements. Among the most important and best-curated databases are Genbank, RefSeq and ENSEMBL. Genbank is an annotated collection of all publicly available DNA sequences (Benson et al., 2013). The RefSeq collection offers a comprehensive, integrated, non-redundant and well-annotated set of sequences, including genomic DNA, transcripts and proteins (Pruitt et al., 2007). The ENSEMBL database provides DNA sequence assemblies and curated Ensembl gene builds from various projects (Yates et al., 2016).

Current meta-genomic pipelines consist of custom-prepared scripts that automatically retrieve selected genomes from these resources. Post-processing, handling and analysis usually uses the Perl, Python, or R programming languages. Although powerful sequence retrieval frameworks have been implemented in Perl (BioPerl), Python (BioPython) and R, their use requires appropriate programming expertise. Furthermore, none of these frameworks combines meta-genomic scale sequence retrieval with functional annotation. These deficiencies also apply to the currently available R packages seqinr and biomaRt. The seqinr package aims to automate sequence retrieval in R but is not designed for meta-genomic approaches and does not include functional annotation. The biomaRt package aims to provide functional annotation methods but these are also not designed for meta-genomic approaches and are not easy to use for non-programming experts. To provide a fast, transparent and easy-to-use framework for combined genomic data retrieval and efficient functional annotation of genetic features in meta-genomic approaches, we have designed the R package biomartr for use with the NCBI, ENSEMBL, ESNEMBLGENOMES and BioMart infrastructures (Smedley et al., 2009). The major advantage of biomartr is that it does not require profound programming expertise. It is optimized to handle multiple genomes simultaneously and allows, for example, assignment of Gene Ontology (GO) information and sequence homology relationships between different organisms by communicating with the BioMart database. The interface functions communicating with the BioMart database use a novel organism centered notation for information retrieval. Instead of learning the underlying database and dataset linking convention of BioMart, users can type the scientific name of an organism of interest (e.g. ‘Homo sapiens’) to retrieve a list of all available information provided by BioMart for this particular organism of interest. In summary, the biomartr package provides researchers with a powerful tool for efficient, straightforward and reproducible handling of large-scale meta-genomic data and intuitive organism centered interface functions to retrieve functional annotation information from the BioMart database.

2 Implementation

The biomartr package is released under the GNU General Public License within the CRAN project (R Core Team). The package can be downloaded from https://cran.r-project.org/web/packages/biomartr/index.html. The source code is publically available at https://github.com/HajkD/biomartr. The biomartr package depends on the R packages Biostrings, data.table, dplyr, readr, downloader, RCurl, XML, biomaRt (Durinck et al., 2005), httr and stringr. The functionality of packages such as biomaRt (Durinck et al., 2005) and seqinr (Charif and Lobry, 2007) are included in biomartr and significantly extended. This is achieved by additional data retrieval functions and the direct combination of biomaRt and seqinr functionality with improved retrieval capability.

3 Functions and examples

Thirty-seven functions are provided by the biomartr package. For genome and database retrieval, the functions listDatabases(), listKingdoms(), listGroups(), listSubgroups(), listGenomes() and is.genome.available() enable the listing of all databases and genomic sequences that are available for automated retrieval.

For example, the entire NCBI nr database can then be downloaded easily using just one command:

download.database.all(db = ‘nr’, path = ‘nr’)

Analogous to the retrieval of databases as described above, selected genomes can also be retrieved using the following function.

As exemplified below by the human genome, download can be triggered by typing:

getGenome (db = ‘refseq’, organism = ‘Homo sapiens’)

The command getGenome() also documents the source and version of the downloaded files. Corresponding download of proteomes, coding sequences and annotation files can be obtained by applying

getProteome(), getCDS() and getGFF(), respectively. The db argument can be specified to retrieve genomes from other NCBI or ENSEMBL databases. For meta-genome approaches, biomartr includes the meta.retrieval() function to download the genomes of entire kingdoms:

# Download all vertebrate genomes

meta.retrieval(kingdom = ‘vertebrate_mammalian’, type = ‘genome’)

Hence, for example, all mammalian genomes can be downloaded with just one command. The type argument can also be specified for proteomes, coding sequences and annotation files. For functional annotation, available datasets and BioMart connections for a specific organism of interest can be obtained by typing:

organismBM(organism = ‘Homo sapiens’)

For example, available sequence homology relationships to other organisms can be retrieved by running the command:

organismAttributes(organism = ‘Homo sapiens’, topic = ‘homolog’)

Finally, users can retrieve GO information for a particular gene or gene set, e.g. human gene GUCA2A by running the command:

getGO(organism = ‘Homo sapiens’,

genes = ‘GUCA2A’,

filters = ‘hgnc_symbol’)

Tutorials are available at https://github.com/HajkD/biomartr#tutorials and also in the Supplementary Tutorial.

4 Conclusions

The functions provided by biomartr enable fast data retrieval and functional annotation queries to prominent sequence and annotation databases such as NCBI, ENSEMBL, ENSEMBLGENOMES and BioMart. In addition, all data retrieval functions implemented in biomartr automatically archive and log the source, date, version, taxid and type of data retrieved. Thus, biomartr improves reproducibility and transparency in genomic data handling. It can be integrated easily into meta-genomic analyses.

Funding

This work was supported by an European Research Council grant named EVOBREED [grant number 322621] (to JP) and a Gatsby Fellowship [grant number AT3273/GLE] (to JP).

Conflict of Interest: none declared.