Sequence_Retrieval.R
In biomartr: Genomic Data Retrieval

## ----echo = FALSE, message = FALSE------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
options(width = 750)
knitr::opts_chunk$set(
  comment = "#>",
  error = FALSE,
  tidy = FALSE)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  length(listGenomes(db = "refseq"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  length(listGenomes(db = "genbank"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  length(listGenomes(db = "ensembl"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # the number of genomes available for each kingdom
#  listKingdoms(db = "refseq")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # the number of genomes available for each kingdom
#  listKingdoms(db = "genbank")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # the number of genomes available for each kingdom
#  listKingdoms(db = "ensembl")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # the number of genomes available for each group
#  listGroups(db = "refseq")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # the number of genomes available for each group
#  listGroups(db = "genbank")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the genome of Homo sapiens from refseq
#  # and store the corresponding genome file in '_ncbi_downloads/genomes'
#  HS.genome.refseq <- getGenome( db       = "refseq",
#                                 organism = "Homo sapiens",
#                                 path     = file.path("_ncbi_downloads","genomes") )

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the genome of Homo sapiens from refseq
#  # and store the corresponding genome file in '_ncbi_downloads/genomes'
#  HS.genome.refseq <- getGenome( db       = "refseq",
#                                 organism = "Homo sapiens")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded genome as Biostrings object
#  Human_Genome <- read_genome(file = HS.genome.refseq)
#  

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # look at the Biostrings object
#  Human_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import genome as Biostrings object
#  Human_Genome <- getGenome( db       = "refseq",
#                             organism = "Homo sapiens",
#                             path     = file.path("_ncbi_downloads","genomes")) %>%
#      read_genome()

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Human_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import genome as Biostrings object
#  Scerevisiae_Genome <- getGenome(
#              db       = "refseq",
#              organism = "559292") %>%
#      read_genome()

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Scerevisiae_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import genome as Biostrings object
#  Scerevisiae_Genome <- getGenome(
#              db       = "refseq",
#              organism = "GCF_000146045.2") %>%
#      read_genome()

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Scerevisiae_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the genome of Homo sapiens from Genbank
#  # and store the corresponding genome file in '_ncbi_downloads/genomes'
#  HS.genome.genbank <- getGenome( db       = "genbank",
#                                  organism = "Homo sapiens",
#                                  path     = file.path("_ncbi_downloads","genomes") )

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded genome as Biostrings object
#  Human_Genome <- read_genome(file = HS.genome.genbank)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # look at the Biostrings object
#  Human_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import genome as Biostrings object
#  Scerevisiae_Genome <- getGenome(
#              db       = "genbank",
#              organism = "559292") %>%
#      read_genome()

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Scerevisiae_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import genome as Biostrings object
#  Scerevisiae_Genome <- getGenome(
#              db       = "genbank",
#              organism = "GCA_000146045.2") %>%
#      read_genome()

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Scerevisiae_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the genome of Homo sapiens from ENSEMBL
#  # and store the corresponding genome file in '_ncbi_downloads/genomes'
#  HS.genome.ensembl <- getGenome( db       = "ensembl",
#                                  organism = "Homo sapiens",
#                                  path     = file.path("_ncbi_downloads","genomes") ,
#                                  assembly_type = "primary_assembly")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded genome as Biostrings object
#  Human_Genome <- read_genome(file = HS.genome.ensembl)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # look at the Biostrings object
#  Human_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import genome as Biostrings object
#  Scerevisiae_Genome <- getGenome(
#              db       = "ensembl",
#              organism = "4932") %>%
#      read_genome()

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Scerevisiae_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import genome as Biostrings object
#  Scerevisiae_Genome <- getGenome(
#              db       = "ensembl",
#              organism = "GCA_000146045.2") %>%
#      read_genome()

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Scerevisiae_Genome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the proteome of Homo sapiens from refseq
#  # and store the corresponding proteome file in '_ncbi_downloads/proteomes'
#  HS.proteome.refseq <- getProteome( db       = "refseq",
#                                     organism = "Homo sapiens",
#                                     path     = file.path("_ncbi_downloads","proteomes"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import proteome as Biostrings object
#  Human_Proteome <- read_proteome(file = HS.proteome.refseq)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Human_Proteome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import proteome as Biostrings object
#  Human_Proteome <- getProteome( db       = "refseq",
#                                 organism = "Homo sapiens",
#                                 path     = file.path("_ncbi_downloads","proteomes")) %>%
#      read_proteome()

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Human_Proteome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the proteome of Homo sapiens from genbank
#  # and store the corresponding proteome file in '_ncbi_downloads/proteomes'
#  HS.proteome.genbank <- getProteome( db       = "genbank",
#                                     organism = "Homo sapiens",
#                                     path     = file.path("_ncbi_downloads","proteomes"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import proteome as Biostrings object
#  Human_Proteome <- read_proteome(file = HS.proteome.genbank)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Human_Proteome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the proteome of Homo sapiens from ENSEMBL
#  # and store the corresponding proteome file in '_ncbi_downloads/proteomes'
#  HS.proteome.ensembl <- getProteome( db       = "ensembl",
#                                     organism = "Homo sapiens",
#                                     path     = file.path("_ncbi_downloads","proteomes"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import proteome as Biostrings object
#  Human_Proteome <- read_proteome(file = HS.proteome.ensembl)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Human_Proteome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the proteome of Mus musculus from UniProt
#  # and store the corresponding proteome file in '_uniprot_downloads/proteomes'
#  Mm.proteome.uniprot<- getProteome( db       = "uniprot",
#                                     organism = "Mus musculus",
#                                     path     = file.path("_uniprot_downloads","proteomes"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import proteome as Biostrings object
#  Mouse_Proteome <- read_proteome(file = Mm.proteome.uniprot)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Mouse_Proteome

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the genome of Homo sapiens from refseq
#  # and store the corresponding genome CDS file in '_ncbi_downloads/CDS'
#  HS.cds.refseq <- getCDS( db       = "refseq",
#                           organism = "Homo sapiens",
#                           path     = file.path("_ncbi_downloads","CDS"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded CDS as Biostrings object
#  Human_CDS <- read_cds(file     = HS.cds.refseq,
#                        obj.type = "Biostrings")
#  

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # look at the Biostrings object
#  Human_CDS

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import CDS as Biostrings object
#  Human_CDS <- getCDS( db       = "refseq",
#                       organism = "Homo sapiens",
#                       path     = file.path("_ncbi_downloads","CDS")) %>%
#      read_cds(obj.type = "Biostrings")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Human_CDS

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the genome of Homo sapiens from genbank
#  # and store the corresponding genome CDS file in '_ncbi_downloads/CDS'
#  HS.cds.genbank <- getCDS( db       = "genbank",
#                           organism = "Homo sapiens",
#                           path     = file.path("_ncbi_downloads","CDS"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded CDS as Biostrings object
#  Human_CDS <- read_cds(file     = HS.cds.genbank,
#                        obj.type = "Biostrings")
#  

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # look at the Biostrings object
#  Human_CDS

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the genome of Homo sapiens from ensembl
#  # and store the corresponding genome CDS file in '_ncbi_downloads/CDS'
#  HS.cds.ensembl <- getCDS( db       = "ensembl",
#                           organism = "Homo sapiens",
#                           path     = file.path("_ncbi_downloads","CDS"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded CDS as Biostrings object
#  Human_CDS <- read_cds(file     = HS.cds.ensembl,
#                        obj.type = "Biostrings")
#  

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # look at the Biostrings object
#  Human_CDS

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the RNA of Homo sapiens from refseq
#  # and store the corresponding RNA file in '_ncbi_downloads/RNA'
#  HS.rna.refseq <- getRNA( db       = "refseq",
#                           organism = "Homo sapiens",
#                           path     = file.path("_ncbi_downloads","RNA"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded RNA as Biostrings object
#  Human_rna <- read_rna(file     = HS.rna.refseq,
#                        obj.type = "Biostrings")
#  

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # look at the Biostrings object
#  Human_rna

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # install.packages("magrittr")
#  library(magrittr)
#  # import RNA as Biostrings object
#  Human_rna <- getRNA( db       = "refseq",
#                       organism = "Homo sapiens",
#                       path     = file.path("_ncbi_downloads","RNA")) %>%
#      read_cds(obj.type = "Biostrings")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Human_rna

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the RNA of Homo sapiens from genbank
#  # and store the corresponding genome RNA file in '_ncbi_downloads/RNA'
#  HS.rna.genbank <- getRNA( db       = "genbank",
#                           organism = "Homo sapiens",
#                           path     = file.path("_ncbi_downloads","RNA"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded RNA as Biostrings object
#  Human_rna <- read_cds(file     = HS.rna.genbank,
#                        obj.type = "Biostrings")
#  

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # look at the Biostrings object
#  Human_rna

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the RNA of Homo sapiens from ensembl
#  # and store the corresponding genome RNA file in '_ncbi_downloads/RNA'
#  HS.rna.ensembl <- getRNA( db       = "ensembl",
#                            organism = "Homo sapiens",
#                            path     = file.path("_ncbi_downloads","RNA"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded RNA as Biostrings object
#  Human_rna <- read_cds(file     = HS.rna.ensembl,
#                        obj.type = "Biostrings")
#  

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # look at the Biostrings object
#  Human_rna

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the GFF file of Homo sapiens from refseq
#  # and store the corresponding file in '_ncbi_downloads/annotation'
#  HS.gff.refseq <- getGFF( db       = "refseq",
#                           organism = "Homo sapiens",
#                           path = file.path("_ncbi_downloads","annotation"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded GFF file
#  Human_GFF <- read_gff(file = HS.gff.refseq)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  Human_GFF

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the GFF file of Homo sapiens from genbank
#  # and store the corresponding file in '_ncbi_downloads/annotation'
#  HS.gff.genbank <- getGFF( db       = "genbank",
#                           organism = "Homo sapiens",
#                           path = file.path("_ncbi_downloads","annotation"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded GFF file
#  Human_GFF <- read_gff(file = HS.gff.genbank)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # show all elements of the data.frame
#  # options(tibble.print_max = Inf)
#  Human_GFF

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the GFF file of Homo sapiens from ENSEMBL
#  # and store the corresponding file in 'ensembl/annotation'
#  HS.gff.ensembl <- getGFF( db       = "ensembl",
#                           organism = "Homo sapiens",
#                           path = file.path("ensembl","annotation"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import downloaded GFF file
#  Human_GFF <- read_gff(file = HS.gff.ensembl)

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # show all elements of the data.frame
#  # options(tibble.print_max = Inf)
#  Human_GFF

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download the GTF file of Homo sapiens from ENSEMBL
#  # and store the corresponding file in 'ensembl/annotation'
#  HS.gtf.ensembl <- getGTF( db       = "ensembl",
#                           organism = "Homo sapiens",
#                           path = file.path("ensembl","annotation"),
#                           assembly_type = "primary_assembly")

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download repeat masker annotation file for Homo sapiens
#  Hsapiens_rm <- getRepeatMasker( db       = "refseq",
#                                   organism = "Homo sapiens",
#                                   path = file.path("refseq","TEannotation"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import TE annotation file
#  Hsapiens_rm_import <- read_rm("refseq/TEannotation/Homo_sapiens_rm_refseq.out.gz")
#  # look at data
#  Hsapiens_rm_import

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download genome assembly stats file for Homo sapiens
#  Hsapiens_stats <- getAssemblyStats( db       = "refseq",
#                                   organism = "Homo sapiens",
#                                   path = file.path("refseq","AssemblyStats"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import TE annotation file
#  Hsapiens_stats_import <- read_assemblystats(Hsapiens_stats)
#  # look at data
#  Hsapiens_stats_import

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download genome assembly stats file for Homo sapiens
#  Hsapiens_stats <- getAssemblyStats( db       = "genbank",
#                                   organism = "Homo sapiens",
#                                   path = file.path("genbank","AssemblyStats"))

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # import TE annotation file
#  Hsapiens_stats_import <- read_assemblystats(Hsapiens_stats)
#  # look at data
#  Hsapiens_stats_import

## ----eval=FALSE-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
#  # download collection for Saccharomyces cerevisiae
#  getCollection( db = "refseq",
#                 organism = "Saccharomyces cerevisiae",
#                 path = file.path("refseq","Collections"))