def delete_sra(self):
"""Delete the downloaded SRA files.
"""
if (pe.deleteFileFromDisk(self.localSRAFilePath)):
del self.localSRAFilePath
return True
return False
def delete_fastq(self):
"""Delte the fastq files from the disk.
The files are referenced by self.localfastqPath or self.localfastq1Path and self.localfastq2Path
"""
if self.layout=='PAIRED':
if pe.deleteMultipleFilesFromDisk(self.localfastq1Path,self.localfastq2Path):
del self.localfastq1Path
del self.localfastq2Path
return True
else:
if pe.deleteFileFromDisk(self.localfastqPath):
del self.localfastqPath
return True
return False
def run_portcullisFull(self,
reference_fasta,
bam_file,
out_dir="",
delete_bam=False,
verbose=False,
quiet=False,
logs=True,
objectid="NA",
**kwargs):
"""
run portculis full
Parameters
----------
reference_fasta: string
Path to the reference fasta file
bam_file: string
Path to input bam file
out_dir: string
Path to the out put dir. current directory is not given.
verbose: bool
Print stdout and std error
quiet: bool
Print nothing
logs: bool
Log this command to pyrpipe logs
objectid: str
Provide an id to attach with this command e.g. the SRR accession. This is useful for debugging, benchmarking and reports.
kwargs: dict
Options to pass to trimgalore. This will override the existing options
"""
if not pu.check_files_exist(reference_fasta, bam_file):
print("Please check input for portcullis.")
return ""
newOpts = {"--": (reference_fasta, bam_file)}
mergedOpts = {**kwargs, **newOpts}
#add out dir path
if not out_dir:
out_dir = os.path.join(os.getcwd(), "portcullis_out")
mergedOpts = {**mergedOpts, **{"-o": out_dir}}
status = self.run_portcullis("full",
verbose=verbose,
quiet=quiet,
logs=logs,
objectid=objectid,
**mergedOpts)
if not status:
print("portcullis full failed for:" + bam_file)
return ""
#check if bam file exists
if not pu.check_paths_exist(out_dir):
return ""
if delete_bam:
if not pe.deleteFileFromDisk(bam_file):
print("Error deleting bam file:" + bam_file)
return out_dir
def sam_to_bam(self,
sam_file,
out_dir="",
out_suffix="",
delete_sam=False,
verbose=False,
quiet=False,
logs=True,
objectid="NA",
**kwargs):
"""Convert sam file to a bam file.
Output bam file will have same name as input sam.
out_suffix: string
Suffix for the output sam file
delete_sam: bool
delete the sam file after conversion
verbose: bool
Print stdout and std error
quiet: bool
Print nothing
logs: bool
Log this command to pyrpipe logs
objectid: str
Provide an id to attach with this command e.g. the SRR accession. This is useful for debugging, benchmarking and reports.
kwargs: dict
Options to pass to trimgalore. This will override the existing options
:return: Returns the path to the bam file. Returns empty string if operation failed.
:rtype: string
"""
if not out_dir:
out_dir = pu.get_file_directory(sam_file)
else:
if not pu.check_paths_exist(out_dir):
pu.mkdir(out_dir)
fname = pu.get_file_basename(sam_file)
#output will be out_bam
out_bam = os.path.join(out_dir, fname + out_suffix + '.bam')
newOpts = {"--": (sam_file, ), "-o": out_bam, "-b": ""}
mergedOpts = {**kwargs, **newOpts}
status = self.run_samtools("view",
verbose=verbose,
quiet=quiet,
logs=logs,
objectid=objectid,
**mergedOpts)
if not status:
print("Sam to bam failed for:" + sam_file)
return ""
#check if bam file exists
if not pu.check_files_exist(out_bam):
return ""
#delete_sam_file
if delete_sam:
if not pe.deleteFileFromDisk(sam_file):
print("Error deleting sam file:" + sam_file)
#return path to file
return out_bam
def merge_bam(self,
*args,
out_file="merged",
out_dir="",
delete_bams=False,
verbose=False,
quiet=False,
logs=True,
objectid="NA",
**kwargs):
"""Merge multiple bam files into a single file
Parameters
----------
out_file: string
Output file name to save the results. .bam will be added at the end.
args:tuple
Paths to bam files to combine
out_dir: string
Path where to save the merged bam file. Default path is the same as the first bam_file's
verbose: bool
Print stdout and std error
quiet: bool
Print nothing
logs: bool
Log this command to pyrpipe logs
objectid: str
Provide an id to attach with this command e.g. the SRR accession. This is useful for debugging, benchmarking and reports.
kwargs: dict
Options to pass to trimgalore. This will override the existing options
:return: Returns the path to the merged bam file.
:rtype: string
"""
if len(args) < 2:
print("Please supply at least 2 files to merge")
return ""
if not out_dir:
out_dir = pu.get_file_directory(args[0])
else:
if not pu.check_paths_exist(out_dir):
pu.mkdir(out_dir)
outMergedFile = os.path.join(out_dir, out_file + ".bam")
newOpts = {"--": (outMergedFile, ) + args}
mergedOpts = {**kwargs, **newOpts}
status = self.run_samtools("merge",
verbose=verbose,
quiet=quiet,
logs=logs,
objectid=objectid,
**mergedOpts)
if not status:
print("Bam merge failed for:" + outMergedFile)
return ""
#check if bam file exists
if not pu.check_files_exist(outMergedFile):
return ""
if delete_bams:
for bam_file in args:
if not pe.deleteFileFromDisk(bam_file):
print("Error deleting sam file:" + bam_file)
return outMergedFile
def sort_bam(self,
bam_file,
out_dir="",
out_suffix="",
threads=None,
delete_bam=False,
verbose=False,
quiet=False,
logs=True,
objectid="NA",
**kwargs):
"""Sorts an input bam file. Outpufile will end in _sorted.bam
bam_file: str
Path to the input bam file
out_dir: str
Path to output directory
out_suffix: str
Output file suffix
threads: int
Number of threads. Default: Use self.threads initialized in init().
delete_bam: bool
Delete input bam_file
verbose: bool
Print stdout and std error
quiet: bool
Print nothing
logs: bool
Log this command to pyrpipe logs
objectid: str
Provide an id to attach with this command e.g. the SRR accession. This is useful for debugging, benchmarking and reports.
kwargs: dict
Options to pass to samtools. This will override the existing options
:return: Returns path to the sorted bam file. Returns empty string if operation failed.
:rtype: string
"""
if not out_dir:
out_dir = pu.get_file_directory(bam_file)
else:
if not pu.check_paths_exist(out_dir):
pu.mkdir(out_dir)
fname = pu.get_file_basename(bam_file)
#output will be out_bam
outSortedbam_file = os.path.join(out_dir,
fname + out_suffix + '_sorted.bam')
#handle threads
if not threads:
threads = self.threads
newOpts = {
"--": (bam_file, ),
"-o": outSortedbam_file,
"-@": str(threads)
}
mergedOpts = {**newOpts, **kwargs}
status = self.run_samtools("sort",
verbose=verbose,
quiet=quiet,
logs=logs,
objectid=objectid,
**mergedOpts)
if not status:
print("Bam sort failed for:" + bam_file)
return ""
#check if bam file exists
if not pu.check_files_exist(outSortedbam_file):
return ""
if delete_bam:
if not pe.deleteFileFromDisk(bam_file):
print("Error deleting sam file:" + bam_file)
#return path to file
return outSortedbam_file
