Chapter 5 Get fastq from SRA database

library(SRAdb) # SRA database
dir.create("~/SRA")
setwd("~/SRA")

## Create database
sqlfile <-'SRAmetadb.sqlite'
# load database [~3 Gb, 60 Gb on disc!!!]
if(!file.exists('SRAmetadb.sqlite')) sqlfile <<- getSRAdbFile()
# connect database
sra_con <- dbConnect(SQLite(), sqlfile)

## get info for SRA from created database
sraInf <- getSRAinfo("SRP045534", sra_con, sraType="sra")
sraInf

# download SRA
# get SRA using SRA info [~ 5 Gb]
sapply(sraInf$run, function(x) try(getSRAfile(x, sra_con, fileType="sra"), silent=TRUE))

5.1 Get fastq from SRA database

library(SRAdb) # SRA database
dir.create("~/SRA")
setwd("~/SRA")

## Create database
sqlfile <-'SRAmetadb.sqlite'
# load database [~3 Gb, 60 Gb on disc!!!]
if(!file.exists('SRAmetadb.sqlite')) sqlfile <<- getSRAdbFile()
# connect database
sra_con <- dbConnect(SQLite(), sqlfile)

## get info for SRA from created database
sraInf <- getSRAinfo("SRP045534", sra_con, sraType="sra")
sraInf

# download SRA
# get SRA using SRA info [~ 5 Gb]
sapply(sraInf$run, function(x) try(getSRAfile(x, sra_con, fileType="sra"), silent=TRUE))

## Alignment using Rsubread

library(Rsubread) # sequence alignment

fastq.files <- list.files(pattern = ".fastq.gz$", full.names = TRUE)
fastq.files

# build index
buildindex(basename="mm10", reference="mm10.fa")

# alignment
align(fastq.files, index="chr1_mm10")

# parameters
args(align)

# check result
bam.files <- list.files(pattern = ".BAM$", full.names = TRUE)
bam.files
# properties of BAM files
propmapped(files=bam.files)