def check_index(self):
"""
Check self.index exists
Returns
-------
bool
True if index exists.
"""
if hasattr(self, 'index'):
return (pu.check_hisatindex(self.index))
else:
return False
def __init__(self, hisat2_index="", **kwargs):
super().__init__()
self.programName = "hisat2"
#check if hisat2 exists
if not pe.check_dependencies([self.programName]):
raise Exception("ERROR: " + self.programName + " not found.")
self.valid_args = [
'-x', '-1', '-2', '-U', '--sra-acc', '-S', '-q', '--qseq', '-f',
'-r', '-c', '-s', '-u', '-5', '-3', '--phred33', '--phred64',
'--int-quals', '--sra-acc', '--n-ceil', '--ignore-quals', '--nofw',
'--norc', '--pen-cansplice', '--pen-noncansplice',
'--pen-canintronlen', '--pen-noncanintronlen', '--min-intronlen',
'--max-intronlen', '--known-splicesite-infile',
'--novel-splicesite-outfile', '--novel-splicesite-infile',
'--no-temp-splicesite', '--no-spliced-alignment',
'--rna-strandness', '--tmo', '--dta', '--dta-cufflinks',
'--avoid-pseudogene', '--no-templatelen-adjustment', '--mp',
'--sp', '--no-softclip', '--np', '--rdg', '--rfg', '--score-min',
'-k', '-I', '-X', '--fr', '--rf', '--ff', '--no-mixed',
'--no-discordant', '-t', '--un', '--al', '--un-conc', '--al-conc',
'--un-gz', '--summary-file', '--new-summary', '--quiet',
'--met-file', '--met-stderr', '--met', '--no-head', '--no-sq',
'--rg-id', '--rgit-sec-seq', '-o', '-p', '--reorder', '--mm',
'--qc-filter', '--seed', '--non-deterministic', '--remove-chrname',
'--add-chrname', '--version'
]
#initialize the passed arguments
self.passedArgumentDict = kwargs
#if index is passed, update the passed arguments
if len(hisat2_index) > 0 and pu.check_hisatindex(hisat2_index):
print("HISAT2 index is: " + hisat2_index)
self.hisat2_index = hisat2_index
self.passedArgumentDict['-x'] = self.hisat2_index
self.index = self.hisat2_index
else:
print(
"No Hisat2 index provided. Please build index now to generate an index using build_Index()...."
)
def build_index(self,
index_path,
index_name,
*args,
threads=None,
overwrite=False,
verbose=False,
quiet=False,
logs=True,
objectid="NA",
**kwargs):
"""Build a hisat index with given parameters and saves the new index to self.hisat2_index.
Parameters
----------
index_path: string
Path where the index will be created
index_name: string
A name for the index
args: tuple
Path to reference input files
threads: int
Num threads to use
verbose : bool
Print stdout and std error
quiet : bool
Print nothing
logs : bool
Log this command to pyrpipe logs
objectid : string
Provide an id to attach with this command e.g. the SRR accession. This is useful for debugging, benchmarking and reports.
kwargs: dict
Parameters for the hisat2-build command
:return: Returns the status of hisat2-build
:rtype: bool
"""
#check input references
if len(args) < 1:
pu.print_boldred(
"No reference sequence provided to hisat2-build. Exiting")
return False
if not pu.check_files_exist(*args):
pu.print_boldred(
"Please check input reference sequences provided to hisat2-build. Exiting"
)
return False
print("Building hisat index...")
hisat2Buildvalid_args = [
'-c', '--large-index', '-a', '-p', '--bmax', '--bmaxdivn', '--dcv',
'--nodc', '-r', '-3', '-o', '-t', '--localoffrate',
'--localftabchars', '--snp', '--haplotype', '--ss', '--exon',
'--seed', '-q', '-h', '--usage', '--version'
]
#create the out dir
if not pu.check_paths_exist(index_path):
if not pu.mkdir(index_path):
print(
"ERROR in building hisat2 index. Failed to create index directory."
)
return False
if not overwrite:
#check if files exists
if pu.check_hisatindex(os.path.join(index_path, index_name)):
print("Hisat2 index with same name already exists. Exiting...")
self.hisat2_index = os.path.join(index_path, index_name)
return True
#handle threads
if not threads:
threads = self.threads
hisat2Build_Cmd = ['hisat2-build']
newOpts = {"-p": str(threads)}
mergedOpts = {**newOpts, **kwargs}
#add options
hisat2Build_Cmd.extend(
pu.parse_unix_args(hisat2Buildvalid_args, mergedOpts))
#add input files
hisat2Build_Cmd.append(str(",".join(args)))
#add dir/basenae
hisat2Build_Cmd.append(os.path.join(index_path, index_name))
#print("Executing:"+str(" ".join(hisat2Build_Cmd)))
#start ececution
status = pe.execute_command(hisat2Build_Cmd,
verbose=verbose,
quiet=quiet,
logs=logs,
objectid=objectid)
if not status:
pu.print_boldred("hisatBuild failed")
return False
#check index files
if not pu.check_hisatindex(os.path.join(index_path, index_name)):
pu.print_boldred("hisatBuild failed")
return False
#set the index path
self.hisat2_index = os.path.join(index_path, index_name)
#return status
return True
def check_index(self):
if hasattr(self, 'hisat2_index'):
return (pu.check_hisatindex(self.hisat2_index))
else:
return False
def build_index(self, index_path, genome, objectid="NA"):
"""Build a hisat index with given parameters and saves the new index to self.index.
Parameters
----------
index_path: string
Path where the index will be created
genome: string
Path to the reference genome
objectid : string
Provide an id to attach with this command e.g. the SRR accession. This is useful for debugging, benchmarking and reports.
:return: Returns the status of hisat2-build
:rtype: bool
"""
#if index already exists then exit
if not _force:
#check if files exists
if pu.check_hisatindex(index_path):
pu.print_green(
"Hisat2 index {} already exists.".format(index_path))
self.index = os.path.join(index_path)
return True
#check input files
if not pu.check_files_exist(genome):
pu.print_boldred(
"Please provide a valid input fasta file to build Hisat2 index"
)
raise ValueError("Please check input to hisat2 build index")
indexdir = pu.get_file_directory(index_path)
#create the out dir
if not pu.check_paths_exist(indexdir):
if not pu.mkdir(indexdir):
raise OSError(
"Error creating hisat2 index. Failed to create index directory."
)
hisat2Buildvalid_args = valid_args._args_HISAT2BUILD
args = (genome, index_path)
internal_kwargs = {"-p": self._threads}
#read build parameters
yamlfile = os.path.join(_params_dir, 'hisat2_index.yaml')
if pu.check_files_exist(yamlfile):
yaml_params = pl.YAML_loader(yamlfile)
yaml_kwargs = yaml_params.get_kwargs()
internal_kwargs = {**yaml_kwargs, **internal_kwargs}
#add positional args
internal_kwargs['--'] = args
hisat2Build_Cmd = ['hisat2-build']
hisat2Build_Cmd.extend(
pu.parse_unix_args(hisat2Buildvalid_args, internal_kwargs))
#execute command
status = pe.execute_command(hisat2Build_Cmd, objectid=objectid)
if status:
if pu.check_hisatindex(index_path) and not _dryrun:
#update object's index
self.index = index_path
if self.check_index():
return True
else:
raise OSError("Error building Hisat2 index")
return True
