# 2019-05-21 biomartr 0.9.0

Please be aware that as of April 2019, ENSEMBLGENOMES was retired (see details here). Hence, all biomartr functions were updated and won’t support data retrieval from ENSEMBLGENOMES servers anymore.

### New Functions

• New function clean.retrieval() enables formatting and automatic unzipping of meta.retrieval output (find out more here: https://ropensci.github.io/biomartr/articles/MetaGenome_Retrieval.html#un-zipping-downloaded-files)
• New function getGenomeSet() allows users to easily retrieve genomes of multiple specified species. In addition, the genome summary statistics for all retrieved species will be stored as well to provide users with insights regarding the genome assembly quality of each species. This file can be used as Supplementary Information file in publications to facilitate reproducible research.
• New function getProteomeSet() allows users to easily retrieve proteomes of multiple specified species
• New function getCDSSet() allows users to easily retrieve coding sequences of multiple specified species
• New function getGFFSet() allows users to easily retrieve GFF annotation files of multiple specified species
• New function getRNASet() allows users to easily retrieve RNA sequences of multiple specified species
• New function summary_genome() allows users to retrieve summary statistics for a genome assembly file to assess the influence of genome assembly qualities when performing comparative genomics tasks
• New function summary_cds() allows users to retrieve summary statistics for a coding sequence (CDS) file. We noticed, that many CDS files stored in NCBI or ENSEMBL databases contain sequences that aren’t divisible by 3 (division into codons). This makes it difficult to divide CDS into codons for e.g. codon alignments or translation into protein sequences. In addition, some CDS files contain a significant amount of sequences that do not start with AUG (start codon). This function enables users to quantify how many of these sequences exist in a downloaded CDS file to process these files according to the analyses at hand.

### New Features of Existing Functions

• the default value of argument reference in meta.retrieval() changed from reference = TRUE to reference = FALSE. This way all genomes (reference AND non-reference) genomes will be downloaded by default. This is what users seem to prefer.
• getCollection() now also retrieves GTF files when db = 'ensembl'
• getAssemblyStats() now also performs md5 checksum test
• all md5 checksum tests now retrieve the new md5checkfile format from NCBI RefSeq and Genbank
• getGTF(): users can now specify the NCBI Taxonomy ID or Accession ID in addition to the scientific name in argument ‘organism’ to retrieve genome assemblies
• getGFF(): users can now specify the NCBI Taxonomy ID or Accession ID for ENSEMBL in addition to the scientific name in argument ‘organism’ to retrieve genome assemblies
• getMarts() will now throw an error when BioMart servers cannot be reached (#36)
• getGenome() now also stores the genome summary statistics (see ?summary_genome()) for the retrieved species in the documentation folder to provide users with insights regarding the genome assembly quality
• In all get*() functions the default for argument reference is now set from reference = TRUE to reference = FALSE (= new default)
• all get*() functions now received a new argument release which allows users to retrieve specific release versions of genomes, proteomes, etc from ENSEMBL and ENSEMBLGENOMES
• all get*() functions received two new arguments clean_retrieval and gunzip which allows users to upzip the downloaded files directly in the get*() function call and rename the file for more convenient downstream analyses

# 2018-06-27 biomartr 0.8.0

### New Functions

• new function getCollection() for retrieval of a collection: the genome sequence, protein sequences, gff files, etc for a particular species

### New Functionality of Existing Functions

• getProteome() can now retrieve proteomes from the UniProt database by specifying getProteome(db = "uniprot"). An example can be found here

• is.genome.available() now prints out more useful interactive messages when searching for available organisms

• is.genome.available() can now handle taxids and assembly_accession ids in addition to the scientific name when specifying argument organism An example can be found here

• is.genome.available() can now check for organism availability in the UniProt database

• getGenome(): users can now specify the NCBI Taxonomy ID or Accession ID in addition to the scientific name in argument ‘organism’ to retrieve genome assemblies

• getProteome(): users can now specify the NCBI Taxonomy ID or Accession ID in addition to the scientific name in argument ‘organism’ to retrieve proteomes

• getCDS(): users can now specify the NCBI Taxonomy ID or Accession ID in addition to the scientific name in argument ‘organism’ to retrieve CDS

• getRNA(): users can now specify the NCBI Taxonomy ID or Accession ID in addition to the scientific name in argument ‘organism’ to retrieve RNAs

• is.genome.available(): argument order was changed from is.genome.available(organism, details, db) to is.genome.available(db, organism, details) to be logically more consistent with all get*() functions
• meta.retrieval receives a new argument restart_at_last to indicate whether or not the download process when re-running the meta.retrieval function shall pick up at the last species or whether it should crawl through all existing files to check the md5checksum
• meta.retrieval now generates an csv overview file in the doc folder which stores genome version, date, origin, etc information for all downloaded organisms and can be directly used as Supplementary Data file in publications to increase computational and biological reproducibility of the genomics study
• download.database.all() can now skip already downloaded files and internally removes corrupted files with non-matching md5checksum. Re-downloading of currupted files and be performed by simply re-running the download.database.all() command

# 2018-01-03 biomartr 0.7.0

### Function changes

• the function meta.retrieval() will now pick up the download at the organism where it left off and will report which species have already been retrieved

• all get*() functions and the meta.retrieval() function receive a new argument reference which allows users to retrieve not-reference or not-representative genome versions when downloading from NCBI RefSeq or NCBI Genbank

• the argument order in meta.retrieval() changed from meta.retrieval(kingdom, group, db, ...) to meta.retrieval(db,kingdom, group, ...) to make the argument order more consistent with the get*() functions

• the argument order in getGroups() changed from getGroups(kingdom, db) to getGroups(db, kingdom) to make the argument order more consistent with the get*() and meta.retrieval() functions

### New Functions

• new internal functions existingOrganisms() and existingOrganisms_ensembl() which check the organisms that have already been downloaded

# biomartr 0.5.1 2017-05-28

### Bug fixes

• fixing a bug in exists.ftp.file() and getENSEMBLGENOMES.Seq() that caused bacterial genome, proteome, etc retrieval to fail due to the wrong construction of a query ftp request https://github.com/HajkD/biomartr/issues/7 (Many thanks to @dbsseven)

• fix a major bug in which organisms having no representative genome would generate NULL paths that subsequently crashed the meta.retrieval() function when it tried to print out the result paths.

### New Functions

• new function getRepeatMasker() for retrieval of Repeat Masker output files

• new function getGTF() for genome annotation retrieval from ensembl and ensemblgenomes in gtf format (Thanks for suggesting it Ge Tan)

• new function getRNA() to perform RNA Sequence Retrieval from NCBI and ENSEMBL databases (Thanks for suggesting it @carlo-berg)

• new function read_rna() for importing Repeat Masker output files downloaded with getRepeatMasker()

• new function read_rm() for importing RNA downloaded with getRNA() as Biostrings or data.table object

• new helper function custom_download() that aims to make the download process more robust and stable -> In detail, the download process is now adapting to the operating system, e.g. using either curl (macOS), wget (Linux), or wininet (Windows)

### Function changes

• function name listDatabases() has been renamed listNCBIDatabases(). In biomartr version 0.6.0 the function name listDatabases() will be depreciated

• meta.retieval() and meta.retieval.all() now allow the bulk retrieval of GTF files for type = 'ensembl' and type = 'esnemblgenomes' via type = "gtf". See getGTF() for more details.

• meta.retieval() and meta.retieval.all() now allow the bulk retrieval of RNA files via type = "rna". See getRNA() for more details.

• meta.retieval() and meta.retieval.all() now allow the bulk retrieval of Repeat Masker output files via type = "rm". See getRepeatMasker() for more details.

• all get*() retrieval functions now skip the download of a particular file if it already exists in the specified file path

• download.database() and download.database.all() now internally perform md5 check sum checks to make sure that the file download was successful

• download.database() and download.database.all() now return the file paths of the downloaded file so that it is easier to use these functions when constructing pipelines, e.g. download.database() %>% ... or download.database.all() %>% ....

• meta.retrieval() and meta.retrieval.all() now return the file paths of the downloaded file so that it is easier to use these functions when constructing pipelines, e.g. meta.retrieval() %>% ... or meta.retrieval() %>% ....

• getGenome(), getProteome(), getCDS(), getRNA(), getGFF(), and getAssemblyStats() now internally perform md5 checksum tests to make sure that files are retrieved intact.

# biomartr 0.4.0 2017-03-14

### Bug fixes

• fixing a major bug https://github.com/HajkD/biomartr/issues/6 that caused that in all get*() (genome, proteome, gff, etc.) and meta.retrieval*() functions the meta retrieval process errored and terminated whenever NCBI or ENSEMBL didn’t store all types of sequences for a particular organism: genome, proteome, cds, etc. This has been fixed now and function calls such as meta.retrieval(kingdom = "bacteria", db = "genbank", type = "proteome") should work properly now (Thanks to @ARamesh123 for making me aware if this bug). Hence, this bug affected all attempts to download all proteome sequences e.g. for bacteria and viruses, because NCBI does not store genome AND proteome information for all bacterial or viral species.

### Function changes

• meta.retrieval() and meta.retrieval.all() can now also download genome assembly stats for all selected species

• meta.retrieval() receives a new argument group that allows users to retrieve species belonging to a subgroup instead of the entire kingdom. Available groups can be retrieved with getGroups().

• functions getSubgroups() and listSubgroups() have been removed and their initial functionality has been merged and integrated into getGroups() and listGroups()

• listGroups() receives a new argument details that allows users to retrieve the organism names that belong to the corresponding subgroups

• getGroups() is now based on listGroups()

• internal function getGENOMESREPORT() is now exported and available to the user

• all organism*() functions now also support Ensembl Plants, Ensembl Metazoa, Ensembl Protist, and Ensembl Fungi (Thanks for pointing out Alex Gabel)

• getMarts() and getDatasets() now also support Ensembl Plants, Ensembl Metazoa, Ensembl Protist, and Ensembl Fungi (Thanks for pointing out Alex Gabel)

• Vignette Meta-Genome Retrieval has more examples how to download genomes of species that belong to the same subgroup

# biomartr 0.3.0 2017-02-09

### Bug fixes

• Fixing a bug https://github.com/HajkD/biomartr/issues/2 based on the readr package that affected the getSummaryFile(), getKingdomAssemblySummary(), getMetaGenomeSummary(), getENSEMBL.Seq() and getENSEMBLGENOMES.Seq() functions causing quoted lines in the assembly_summary.txt to be omitted when reading these files. This artefact caused that e.g. instead of information of 80,000 Bacteria genomes only 40,000 (which non-quotations) were read (Thanks to Xin Wu).

# biomartr 0.2.1 2016-12-15

In this version of biomartr the organism*() functions were adapted to the new ENSEMBL 87 release in which organism name specification in the Biomart description column was changed from a scientific name convention to a mix of common name and scientific name convention.

• all organism*() functions have been adapted to the new ENSEMBL 87 release organism name notation that is used in the Biomart description

• fixing error handling bug that caused commands such as download.database(db = "nr.27.tar.gz") to not execute properly

# biomartr 0.2.0 2016-11-22

In this version, biomartr was extended to now retrieve genome, proteome, CDS, GFF and meta-genome data also from ENSEMBL and ENSEMLGENOMES. Furthermore, all NCBI retrieval functions were updated to the new server folder structure standards of NCBI.

### New Functions

• new meta-retrieval function meta.retrieval.all() allows users to download all individual genomes of all kingdoms of life with one command

• new metagenome retrieval function getMetaGenomes() allows users to retrieve metagenome projects from NCBI Genbank

• new metagenome retrieval function getMetaGenomeAnnotations() allows users to retrieve annotation files for genomes belonging to a metagenome project stored at NCBI Genbank

• new retrieval function getGFF() allows users to retrieve annotation (*.gff) files for specific genomes from NCBI and ENSEMBL databases

• new import function read_gff() allowing users to import GFF files downloaded with getGFF()

• new internal functions to check for availability of ENSEMBL or ENSEMBLGENOMES databases

• new database retrieval function download.database.all() allows users to download entire NCBI databases with one command

• new function listMetaGenomes() allowing users to list available metagenomes on NCBI Genbank

• new external helper function getSummaryFile() to retrieve the assembly_summary.txt file from NCBI

• new external helper function getKingdomAssemblySummary() to retrieve the assembly_summary.txt files from NCBI for all kingdoms and combine them into one big data.frame

• new function listKingdoms() allows users to list the number of available species per kingdom of life

• new function listGroups() allows users to list the number of available species per group

• new function listSubgroups() allows users to list the number of available species per subgroup

• new function getGroups() allows users to retrieve available groups for a kingdom of life

• new function getSubgroups() allows users to retrieve available subgroups for a kingdom of life

• new external helper function getMetaGenomeSummary() to retrieve the assembly_summary.txt files from NCBI genbank metagenomes

• new internal helper function getENSEMBL.Seq() acting as main interface function to communicate with the ENSEMBL database API for sequence retrieval

• new internal helper function getENSEMBLGENOMES.Seq() acting as main interface function to communicate with the ENSEMBL database API for sequence retrieval

• new internal helper function getENSEMBL.Annotation() acting as main interface function to communicate with the ENSEMBL database API for GFF retrieval

• new internal helper function getENSEMBLGENOMES.Annotation() acting as main interface function to communicate with the ENSEMBL database API for GFF retrieval

• new internal helper function get.ensemblgenome.info() to retrieve general organism information from ENSEMBLGENOMES

• new internal helper function get.ensembl.info() to retrieve general organism information from ENSEMBL

• new internal helper function getGENOMEREPORT() to retrieve the genome reports file from ftp://ftp.ncbi.nlm.nih.gov/genomes/GENOME_REPORTS/overview.txt

• new internal helper function connected.to.internet() enabling internet connection check

### Function changes

• functions getGenome(), getProteome(), and getCDS() now can also in addition to NCBI retrieve genomes, proteomes or CDS from ENSEMBL and ENSEMLGENOMES

• the functions getGenome(), getProteome(), and getCDS() were completely re-written and now use the assembly_summary.txt files provided by NCBI to retrieve the download path to the corresponding genome. Furthermore, these functions now lost the kingdom argument. Users now only need to specify the organism name and not the kingdom anymore. Furthermore, all get* functions now return the path to the downloaded genome so that this path can be used as input to all read_* functions.

• download_databases() has been renamed to download.databases() to be more consistent with other function notation

• the argument db_format was removed from listDatabases() and download.database() because it was misleading

• the command listDatabases("all") now returns all available NCBI databases that can be retrieved with download.database()

• download.database() now internally checks if input database specified by the user is actually available on NCBI servers

• the documentary file generated by getGenome(), getProteome(), and getCDS() is now extended to store more details about the downloaded genome

• argument database in is.genome.available() and listGenomes() has been renamed to db to be consistent with all other sequence retrieval functions

• is.genome.available() now also checks availability of organisms in ENSEMBL. See db = "ensembl"

• the argument db_name in listDatabases() has been renamed db to be more consistent with the notation in other functions

• the argument name in download.database() has been renamed db to be more consistent with the notation in other functions

• getKingdoms() now retrieves also kingdom information for ENSEMBL and ENSEMBLGENOMES

• getKingdoms() received new argument db to specify from which database (e.g. refseq, genbank, ensembl or ensemblgenomes) kingdom information shall be retrieved

• getKingdoms(db = "refseq") received one more member: "viral", allowing the genome retrieval of all viruses

• argument out.folder in meta.retrieval() has been renamed to path to be more consistent with other retrieval functions

• all read_* functions now received a new argument obj.type allowing users to choose between storing input genomes as Biostrings object or data.table object

• all read_* functions now have format = "fasta" as default

• the kingdom argument in the listGenomes() function was renamed to type, now allowing users to specify not only specify kingdoms, but also groups and subgroups. Use: listGenomes(type = "kingdom") or listGenomes(type = "group") or listGenomes(type = "subgroup")

• the listGenomes() function receives a new argument subset to specify a subset of the selected type argument. E.g. subset = "Eukaryota" when specifying type = "kingdom"

• new Vignette Meta-Genome Retrieval
• Update examples and extend Introduction Vignette
• Update examples and extend Database Retrieval Vignette
• Update examples and extend Sequence Retrieval Vignette
• Update examples and extend Functional Annotation Vignette

# biomartr 0.1.0 2016-08-07

• fixing a parsing error of the file ftp://ftp.ncbi.nlm.nih.gov/genomes/refseq/vertebrate_mammalian/assembly_summary.txt The problem was that comment lines were introduced and columns couldn’t be parsed correctly anymore. This caused that genomes, proteomes, and CDS files could not be downloaded properly. This has been fixed now.

• genomes, proteome, and CDS as well as meta-genomes can now be retrieved from RefSeq and Genbank (not only RefSeq); only getCDS() does not have genebank access, becasue genbank does not provide CDS sequences

• adding new function meta.retrieval() to mass retrieve genomes for entire kingdoms of life

• fixed a major bug in organismBM() causing the function to fail. The failure of this function affected all downstream organism*() functions. Bug is now fixed and everything works properly

• updated Vignettes

# biomartr 0.0.3 2016-03-02

• updating unit tests for new API

• fixing API problems that caused all BioMart related functions to fail

• fixing retrieval problems in getCDS(), getProteome(), and getGenome()

• the listDatabases() function now has a new option db_name = "all" allowing users to list all available databases stored on NCBI

### Vignettes

• adding new vignette: Database Retrieval
• update the vignettes: Phylotranscriptomics, Sequence Retrieval, and Functional Annotation

# biomartr 0.0.2 2015-08-07

### Vignettes

• adding vignettes: Introduction, Functional Annotation, Phylotranscriptomics, and Sequence Retrieval

Release Version