def __init__(self):
if Config.ini:
self.fh = Config.ini
else:
self.fh = get_path() + '/config/sv2.ini'
self.json = get_path() + '/config/sv2_clf.json'
self.fasta_chr_flag = False
self.clfs = None
self.gtclf = {} # classifiers for genotyping
self.gcfh = get_path() + '/resources/GC_content_reference.txt'
self.gc_dict = {}
self.resource = None # directory to resource files
if not os.path.isfile(self.fh):
print 'FATAL ERROR: {} config file is missing! Please reinstall sv2: [$ pip uninstall sv2 -y && pip install sv2-VERSION.tar.gz]'.format(
self.fh)
sys.stderr.write(
'FATAL ERROR: {} config file is missing! Please reinstall sv2: [$ pip uninstall sv2 -y && pip install sv2-VERSION.tar.gz]\n'
.format(self.fh))
sys.exit(1)
if not os.path.isfile(self.json):
print 'FATAL ERROR: {} json file is missing! Please reinstall sv2: [$ pip uninstall sv2 -y && pip install sv2-VERSION.tar.gz]'.format(
self.json)
sys.stderr.write(
'FATAL ERROR: {} json file is missing! Please reinstall sv2: [$ pip uninstall sv2 -y && pip install sv2-VERSION.tar.gz]\n'
.format(self.json))
sys.exit(1)
if not os.path.isfile(self.gcfh):
print 'FATAL ERROR: {} file is missing! Please reinstall sv2: [$ pip uninstall sv2 -y && pip install sv2-VERSION.tar.gz]'.format(
self.gcfh)
sys.stderr.write(
'FATAL ERROR: {} file is missing! Please reinstall sv2: [$ pip uninstall sv2 -y && pip install sv2-VERSION.tar.gz]\n'
.format(self.gcfh))
sys.exit(1)
with open(self.json) as data_file:
self.clfs = json.load(data_file)
def load_clf(self, jsonfh=None):
import shutil
errFH(jsonfh)
clfs = {}
with open(jsonfh) as f:
clfs = json.load(f)
realpaths = []
for name in clfs:
print 'loading {} classifier ...'.format(name)
clf_dir = get_path() + '/resources/training_sets/' + name + '/'
if not os.path.exists(clf_dir): os.makedirs(clf_dir)
for x in clfs[name]:
clffh = str(clfs[name][x])
if not os.path.isfile(clffh):
sys.stderr.write(
'WARNING: {} does not exist. Please check the paths in {}\n'
.format(clffh, jsonfh))
else:
clfname = clffh.split('/').pop()
newpath = clf_dir + clfname
clfreplace = True
if os.path.isfile(newpath):
clfreplace = query_yes_no(
'WARNING: {} exists... Replace?'.format(newpath),
'no')
if clfreplace == True:
if self.clfs.get(name) == None: self.clfs[name] = {}
self.clfs[name][x] = newpath
realpaths.append(newpath)
shutil.copyfile(clffh, newpath)
for x in realpaths:
print 'installed classifier {}'.format(x)
print 'appending to {}... DO NOT ALTER {}'.format(self.json, self.json)
dump_json(self.json, self.clfs)
def download_resource(self, ):
install_loc = get_path() + '/resources/'
change_loc = query_yes_no(
'Install sv2 resource files to the default location? < {} >'.
format(install_loc))
if change_loc == False:
sys.stdout.write('Enter sv2 resource file destination:\n')
install_loc = raw_input()
if not install_loc.endswith('/'): install_loc = install_loc + '/'
if not os.path.isdir(install_loc):
sys.stderr.write(
'FATAL ERROR: {} not a directory\n'.format(install_loc))
sys.stdout.write(
'FATAL ERROR: {} not a directory\n'.format(install_loc))
sys.exit(1)
install_ok = query_yes_no(
'sv2 resource files will require {} of free space in {}. Proceed with installation?'
.format(__memsize__, install_loc))
if install_ok == False: sys.exit(1)
else:
zip_file = install_loc + 'sv2_resources.zip'
wget.download(__resource_url__, zip_file)
zip_fh = zipfile.ZipFile(zip_file)
if zip_fh.testzip() is not None:
sys.stdout.write(
'FATAL ERROR: {} corrupted. Try <sv2 -download> again\n'.
format(zip_file))
sys.stderr.write(
'FATAL ERROR: {} corrupted. Try <sv2 -download> again\n'.
format(zip_file))
sys.exit(1)
zip_fh.extractall(install_loc)
zip_fh.close()
os.remove(zip_file)
self.resource = install_loc
self.write_config(None, None, None)
sys.stdout.write(
'INSTALL COMPLETE\nRun < sv2 -hg19 FASTA [-hg38,-mm10] > to configure sv2\n'
)
def write_config(self, hg19=None, hg38=None, mm10=None):
conf = ConfigParser()
for x in self.clfs['default']:
clf_fh = str(self.clfs['default'][x])
if not os.path.isfile(clf_fh):
realpath = get_path() + '/resources/training_sets/' + clf_fh
if os.path.isfile(realpath): self.clfs['default'][x] = realpath
else:
print 'FATAL ERROR: {} pickle file not found. If this file is missing, reinstall sv2: pip uninstall sv2 -y && pip install sv2-VERSION.tar.gz'.format(
realpath)
sys.stderr.write(
'FATAL ERROR: {} pickle file not found. If this file is missing, reinstall sv2: pip uninstall sv2 -y && pip install sv2-VERSION.tar.gz\n'
.format(realpath))
sys.exit(1)
dump_json(self.json, self.clfs)
if not os.path.isfile(self.fh):
conf_fh = open(self.fh, 'w')
conf.add_section('FASTA_PATHS')
conf.set('FASTA_PATHS', 'hg19', hg19)
conf.set('FASTA_PATHS', 'hg38', hg38)
conf.set('FASTA_PATHS', 'mm10', mm10)
conf.set('RESOURCE_DIR', 'sv2_resource', self.resource)
conf.write(conf_fh)
conf_fh.close()
else:
conf.read(self.fh)
if hg19 != None:
errFH(hg19)
conf.set('FASTA_PATHS', 'hg19', hg19)
if hg38 != None:
errFH(hg38)
conf.set('FASTA_PATHS', 'hg38', hg38)
if mm10 != None:
errFH(mm10)
conf.set('FASTA_PATHS', 'mm10', mm10)
if self.resource != None:
conf.set('RESOURCE_DIR', 'sv2_resource', self.resource)
with open(self.fh, 'w') as conf_fh:
conf.write(conf_fh)