Use Genomic Ranges with Rice data

download_ensembl.sh

curl -O ftp://ftp.ensemblgenomes.org/pub/release-46/plants/gff3/oryza_sativa/Oryza_sativa.IRGSP-1.0.46.gff3.gz
curl -O ftp://ftp.ensemblgenomes.org/pub/release-46/plants/gtf/oryza_sativa/Oryza_sativa.IRGSP-1.0.46.chr.gtf.gz
curl -O ftp://ftp.ensemblgenomes.org/pub/release-46/plants/fasta/oryza_sativa/dna/Oryza_sativa.IRGSP-1.0.dna_sm.toplevel.fa.gz

Rice_ranges.R

library(GenomicFeatures)
if( file.exists("MSU_7.db")){
  txdb = loadDb("MSU_7.db")
} else {
  txdb <- makeTxDbFromGFF("/bigdata/wesslerlab/shared/Rice/GFF/MSU_r7/MSU_r7.all.gff3",
                          dataSource="MSU7",
                          organism="Oryza sativa")
  saveDb(txdb,"MSU_7.db")
}
ebg <- exonsBy(txdb, by="gene")
gene_lengths = sum(width(reduce(ebg)))
summary(gene_lengths)

if( file.exists("IRGSP-1.0.db")){
  txdb = loadDb("IRGSP-1.0.db")
} else {
  # can read from .gzip compressed also
  txdb <- makeTxDbFromGFF("/bigdata/wesslerlab/shared/Rice/GFF/IRGSP-1.0/Oryza_sativa.IRGSP-1.0.46.gff3.gz",
                          dataSource="Ensembl",
                          organism="Oryza sativa")
  saveDb(txdb,"IRGSP-1.0.db")
}

ebg <- exonsBy(txdb, by="gene")
gene_lengths = sum(width(reduce(ebg)))
summary(gene_lengths)

if( file.exists("IRGSP-1.0_gtf.db")){
  txdb = loadDb("IRGSP-1.0_gtf.db")
} else {
  
  txdb <- makeTxDbFromGFF("/bigdata/wesslerlab/shared/Rice/GFF/IRGSP-1.0/Oryza_sativa.IRGSP-1.0.46.chr.gtf",
                          dataSource="Ensembl",
                          organism="Oryza sativa")
  saveDb(txdb,"IRGSP-1.0_gtf.db")
}

ebg <- exonsBy(txdb, by="gene")
gene_lengths = sum(width(reduce(ebg)))
summary(gene_lengths)