def search_sra(self, path):
"""Search .sra file under a dir
Return True if found otherwise False
"""
#search files under the path
sra_files = pe.find_files(path, "*.sra")
if len(sra_files) < 1:
return False
if len(sra_files) > 1:
pu.print_boldred(
"Found multiple .sra files. Using the first entry...")
sra_path = sra_files[0]
#self.location=path
self.srr_accession = pu.get_file_basename(sra_path)
self.localSRAFilePath = sra_path
self.sraFileSize = pu.get_file_size(self.localSRAFilePath)
#test if file is paired or single end
if pe.is_paired(self.localSRAFilePath):
self.layout = "PAIRED"
else:
self.layout = "SINGLE"
pu.print_green("Found .sra " + self.localSRAFilePath)
return True
def init_from_accession(self, srr_accession, location):
"""Create SRA object using provided srr accession and location to save the data
"""
self.dep_list = ['prefetch', "fasterq-dump"]
if not pe.check_dependencies(self.dep_list):
raise Exception("ERROR: Please install missing programs.")
if srr_accession is None:
raise Exception("Please provide a valid accession")
if location is None:
location = os.getcwd()
#pu.print_info("Creating SRA: "+srr_accession)
self.srr_accession = srr_accession
#create a dir named <srr_accession> and use as location
self.location = os.path.join(location, self.srr_accession)
#search for existing files in location
#self.search_fastq(self.location)
#scan path for sra
#self.search_sra(self.location)
#check SRA file
if pu.check_files_exist(
os.path.join(self.location, self.srr_accession + ".sra")):
pu.print_green(self.srr_accession + ".sra exists.")
self.localSRAFilePath = os.path.join(self.location,
self.srr_accession + ".sra")
self.sraFileSize = pu.get_file_size(self.localSRAFilePath)
#test if file is paired or single end
if pe.is_paired(self.localSRAFilePath):
self.layout = "PAIRED"
else:
self.layout = "SINGLE"
#check fastq file
self.search_fastq(self.location)
def download_sra(self, verbose=False, quiet=False, logs=True, **kwargs):
"""This function downloads .sra file from NCBI SRA servers using the prefetch command.
NCBI sra-toolkit 2.9 or higher must be installed on the system in order to use prefetch.
prefetch will create a folder with name same as <srr_accession> under the location (path) specified.
The path of downloaded file is saved in the object as localSRAPath. This localSRAPath is then used
by other functions to access the downloaded data.
The **kwargs is for passing arguments to the prefetch command.
Parameters
----------
kwargs: dict
dict containing additional prefetch arguments
:return: Return status of the prefetch command. True if successful download and False if failed.
:rtype: bool
Examples
--------
>>> object.download_sra()
True
"""
#store path to the downloaded sra file
self.localSRAFilePath = os.path.join(self.location,
self.srr_accession + ".sra")
#check if already exists
if pu.check_files_exist(self.localSRAFilePath):
pu.print_green("File already exists:" + self.localSRAFilePath)
#save file .sra file size
self.sraFileSize = pu.get_file_size(self.localSRAFilePath)
#test if file is paired or single end
if pe.is_paired(self.localSRAFilePath):
self.layout = "PAIRED"
else:
self.layout = "SINGLE"
return True
pu.print_info("Downloading " + self.srr_accession + " ...")
#scan for prefetch arguments
prefetchArgsList = [
'-f', '-t', '-l', '-n', '-s', '-R', '-N', '-X', '-o', '-a',
'--ascp-options', '-p', '--eliminate-quals', '-c', '-o', '-O',
'-h', '-V', '-L', '-v', '-q'
]
#ignore location and file name arguments if given
if '-O' in kwargs:
print("Ignoring -O flag." + " location is: " + self.location)
#delete -O parameter
del kwargs['-O']
if '-o' in kwargs:
print("Ignoring -o flag." + " File name is: " + self.srr_accession)
#delete -o parameter
del kwargs['-o']
prefetch_Cmd = ['prefetch']
prefetch_Cmd.extend(pu.parse_unix_args(prefetchArgsList, kwargs))
prefetch_Cmd.extend(['-O', self.location])
prefetch_Cmd.append(self.srr_accession)
cmdStatus = pe.execute_command(prefetch_Cmd,
objectid=self.srr_accession)
if not cmdStatus:
pu.print_boldred("prefetch failed for:" + self.srr_accession)
return False
#validate path exists
if not pu.check_files_exist(self.localSRAFilePath):
pu.print_boldred("Error downloading file. File " +
self.localSRAFilePath + " does not exist!!!")
return False
print("Downloaded file: " + self.localSRAFilePath +
" {0} ".format(pu.get_file_size(self.localSRAFilePath)))
#save file .sra file size
self.sraFileSize = pu.get_file_size(self.localSRAFilePath)
#test if file is paired or single end
if pe.is_paired(self.localSRAFilePath):
self.layout = "PAIRED"
else:
self.layout = "SINGLE"
return True