def createMikadoGTFlist(self,
out_file,
out_dir,
searchPath,
searchQuery="*.gtf",
strand=False):
"""Create a file to be used by mikado configure
"""
files = pe.find_files(searchPath, searchQuery)
args = files
#create out dir
if not pu.check_paths_exist(out_dir):
pu.mkdir(out_dir)
outFilePath = os.path.join(out_dir, out_file + ".txt")
gtfs = []
for l in args:
thisName = pu.get_file_basename(l)
if thisName:
gtfs.append("\t".join([l, thisName, str(strand)]))
f = open(outFilePath, "w")
f.write("\n".join(gtfs))
f.close()
pu.print_green("Mikado list file written to:" + outFilePath)
return outFilePath
def search_fastq(self, path):
"""Search .fastq file under a dir and create SRA object
Return True if found otherwise False
"""
#search files under the path
fq_files = pe.find_files(path, "*.fastq")
if len(fq_files) < 1:
return False
if len(fq_files) > 2:
pu.print_boldred("Can not determine .fastq. Exiting...")
return False
fq_files.sort()
#case with single fastq
if len(fq_files) == 1:
self.localfastqPath = fq_files[0]
pu.print_green("Found .fastq " + self.localfastqPath)
self.layout = "SINGLE"
#case with paired fastq
if len(fq_files) == 2:
self.localfastq1Path = fq_files[0]
self.localfastq2Path = fq_files[1]
pu.print_green("Found .fastq " + self.localfastq1Path + " " +
self.localfastq2Path)
self.layout = "PAIRED"
#self.location=path
#self.srr_accession=pu.get_file_basename(fq_files[0])
return True
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 createMikadoGTFlist(self,
out_file,
out_dir,
searchPath,
searchQuery="*.gtf",
strand=False):
"""Create a file to be used by mikado configure
out_file: str
outfile name
out_dir: str
path to out_dir
searchPath: str
Path where gtf/gff files will be searched
searchQuery: str
Query to perform search. Default: "*.gtf"
strand: bool
Stranded flag: Default false
"""
files = pe.find_files(searchPath, searchQuery, recursive=True)
args = files
#create out dir
if not pu.check_paths_exist(out_dir):
pu.mkdir(out_dir)
outFilePath = os.path.join(out_dir, out_file + ".txt")
gtfs = []
for l in args:
thisName = pu.get_file_basename(l)
if thisName:
gtfs.append("\t".join([l, thisName, str(strand)]))
f = open(outFilePath, "w")
f.write("\n".join(gtfs))
f.close()
pu.print_green("Mikado list file written to:" + outFilePath)
return outFilePath
def download_fastq(self,
verbose=False,
quiet=False,
logs=True,
procs=2,
**kwargs):
"""Function to download fastq files
"""
#check if fastq files exists already
if self.fastqFilesExistsLocally():
pu.print_green("Fastq files exist already")
return True
fasterqdumpArgsList = [
'-f', '-t', '-s', '-N', '-X', '-a', '-p', '-c', '-o', '-O', '-h',
'-V', '-L', '-v', '-q', '-b', '-m', '-x', '-S', '-3', '-P', '-M',
'-B', '--option-file', '--strict', '--table',
'--include-technical', '--skip-technical', '--concatenate-reads'
]
fstrqd_Cmd = ['fasterq-dump']
fstrqd_Cmd.extend(pu.parse_unix_args(fasterqdumpArgsList, kwargs))
#add location
fstrqd_Cmd.extend(['-O', self.location])
#add output filename. output will be <srr_accession>.fastq or <srr_accession>_1.fastq and <srr_accession>_2.fastq
fstrqd_Cmd.extend(['-o', self.srr_accession + ".fastq"])
fstrqd_Cmd.extend(['-e', str(procs)])
if self.sraFileExistsLocally():
fstrqd_Cmd.append(self.localSRAFilePath)
else:
fstrqd_Cmd.append(self.srr_accession)
#execute command
cmdStatus = pe.execute_command(fstrqd_Cmd, objectid=self.srr_accession)
if not cmdStatus:
print("fasterqdump failed for:" + self.srr_accession)
return False
if not hasattr(self, 'layout'):
fq_files = pe.find_files(self.location,
self.srr_accession + "*.fastq")
if len(fq_files) == 1:
self.layout = 'SINGLE'
else:
self.layout = 'PAIRED'
#check if fastq files are downloaded
if (self.layout == "SINGLE"):
self.localfastqPath = os.path.join(self.location,
self.srr_accession + ".fastq")
if not pu.check_files_exist(self.localfastqPath):
pu.print_boldred("Error running fasterq-dump file. File " +
self.localfastqPath + " does not exist!!!")
return False
else:
self.localfastq1Path = os.path.join(
self.location, self.srr_accession + "_1.fastq")
self.localfastq2Path = os.path.join(
self.location, self.srr_accession + "_2.fastq")
if not pu.check_files_exist(self.localfastq1Path,
self.localfastq2Path):
pu.print_boldred("Error running fasterq-dump file. File " +
self.localfastq1Path + " does not exist!!!")
return False
return True