def get_build_num(self, dir=None, is_ret_int=True):
if dir is None:
raise ValueError('dir is empty.')
version_info = VersionManager(dir)
if is_ret_int:
build_num = int(version_info.build)
else:
build_num = version_info.build
return build_num
def merge(self):
fv_tissue = list()
# build merged fv list by tissue: fv_tissue
for fv_set in self.fv:
for idx, file in enumerate(fv_set):
if len(fv_tissue) <= idx:
fv_tissue.append([file])
else:
fv_tissue[idx].append(file)
# merge feature vectors' contents
for files in fv_tissue: # for all 23 tissues
merged = list()
tissue = None
label_name = ''
# step 1: set fv contents with list() * dict()
fv_dicts = list() # list() * dict() * len(files)
for file in files:
file_path = file.path + file.file_name
tmp_dict = CSVLoader.csv2dict(filename=file_path)
fv_dicts.append(tmp_dict)
if tissue is None:
tissue = int(file.tissue)
build_num = VersionManager(file.version).build
label_name += '-' + file.label_type + file.label_name + '_' + str(build_num)
# step 2: merge
# 2.1 get common gnids
common_gnids = list()
for fv_dict in fv_dicts:
gnids = list(fv_dict)
if len(common_gnids) == 0:
common_gnids = gnids.copy()
else:
common_gnids = ListTool.common_items(common_gnids, gnids)
# 2.2 merge
for gnid in common_gnids:
fv_m = list()
for fv_dict in fv_dicts:
fv_m += fv_dict[gnid][:-1]
new_row = [str(gnid)] + fv_m + ['1' if sum([int(fv[gnid][-1]) for fv in fv_dicts]) >= 2 else '0']
#new_row = [str(gnid)] + fv_m + ['1' if sum([int(fv[gnid][-1]) for fv in fv_dicts]) >= 1 else '0']
merged.append(new_row)
# show total number of gnids and possitive class
print('total gnids: {}\tpositive class#: {}'.format(len(merged), sum([int(x[-1]) for x in merged])))
# set FvFile with merged FV
file_name = 'maize_gp_mt%02d%s_%s.arff' % (tissue, label_name, settings.DEV_VERSION)
path = settings.RESULT_DIR + settings.DEV_VERSION + '/0/1/'
self.fv_merged.append(FvFile(file_name=file_name, path=path, data=merged))
print('\t\t===> ' + path + file_name)
def main(argv):
src = None
dst = None
k_size = None
args = parser.parse_args(argv[1:])
if len(argv) <= 1:
parser.parse_args(['--help'])
return
## get kmer size
#if args.ksize:
# k_size = args.ksize
# get source dir
if args.src:
src = args.src
if not os.path.exists(src):
raise FileNotFoundError('{} does not exist.'.format(src))
# set target dir
# get destination dir
if args.dst:
dst = args.dst
else:
dst = src + '../0_arffs/' # default
print('version:', settings.DEV_VERSION)
print("source directory:", src)
print("destination directory:", dst)
#
# Rename and aggregating to the destination dir
#
all_dirs_info = FvManager.get_feature_dirs(src)
all_dirs_info.sort()
for dir in all_dirs_info:
version_info = VersionManager(version_str=dir)
src_arffs_dir = src + dir + '/0/1/'
files_list = FileList.ls(path=src_arffs_dir,
recursion=False,
mode=settings.FLS_FILE_ONLY,
ext=settings.FLS_EXT_ARFF)
for file in files_list:
if file.get_ext().lower() == settings.FLS_EXT_ARFF:
#print(file.full)
new_file = dst + file.name
FileManager.dir_create(dst)
copyfile(src=file.full, dst=new_file)
print(new_file)
def create_arff(self):
# get feature names
feature_names = list()
version = VersionManager(version_str=settings.DEV_VERSION)
for feature in self.features:
feature_names.append(feature.name.strip())
# write arff file for each feature
for feature in self.features:
file_name = 'maize_gp_{BD}_K{KM}_{FN}'.format(
BD=version.build, KM=self.kmer_size, FN=feature.name.strip())
file_path = '{RD}{VER}/{IT}/{RS}/{FI}.arff'.format(
RD=settings.RESULT_DIR,
VER=settings.DEV_VERSION,
KM=self.kmer_size,
IT=self.debug_mode[1],
RS=self.debug_mode[0],
FI=file_name)
f = FileManager.file_open(file_path, 'w')
# arff header
#f.write('@relation maize-gp-%s\n' % feature.name.strip())
f.write('@relation %s\n' % file_name)
for feature_attr in feature_names:
f.write('@attribute %s numeric\n' % feature_attr)
f.write('@attribute class {1,0}\n')
f.write('@data\n')
# arff data
# gnids
gnids = self.exp_setting.get_gene_dataset_gnids_list(
feature_id=feature.corresp_tissue)
for gnid in gnids:
vector = self.feature_vector.get(gnid, None)
if vector is not None:
line = ",".join(str(value) for value in vector)
predicted_results = feature.prediction_results.get(gnid)
if predicted_results is None:
data_label = '?'
else:
data_label = predicted_results.get_assigned_class()
f.write("%s,%s\n" % (line, data_label))
f.close()
def main(argv):
builds_list = None
source_dir_path = None
main_exp_num = None
seq_type = None
args = parser.parse_args(argv[1:])
if len(argv) <= 1:
parser.parse_args(['--help'])
return
# get seq type
if args.seq_type:
seq_type_str = args.seq_type
if seq_type_str.lower() in ['p', 'protein', 'peptide']:
seq_type = 'Protein'
elif seq_type_str.lower() in ['d', 'dna']:
seq_type = 'DNA'
elif seq_type_str.lower() in ['m', 'promoter', 'pmt', 'pm']:
seq_type = 'Promoter'
elif seq_type_str.lower() in ['r', 'rda', 'ra']:
seq_type = 'Reduced Alphabet'
else:
seq_type = 'Unknown - ' + seq_type_str
print('seq type: {}'.format(seq_type))
# get features number to combine
if args.builds:
arg_str = args.builds[0]
print(arg_str)
if arg_str.lower() == 'all':
builds_list = None
else:
builds_list = StrTool.ranges2list(arg_str)
else:
builds_list = None # default: all
print('Builds: {}'.format(builds_list))
# get source directory
if args.src:
source_dir_path = args.src
main_exp_num = source_dir_path[-20:-11]
if not os.path.exists(source_dir_path):
raise FileNotFoundError(
'{} does not exist.'.format(source_dir_path))
print('source directory:', source_dir_path)
#
# build aggregated results summary
#
aggregated_summary = None
exp_num = None
all_dirs_info = FvManager.get_feature_dirs(source_dir_path)
all_dirs_info.sort()
for dir in all_dirs_info:
version_info = VersionManager(version_str=dir)
if builds_list is None or version_info.build in builds_list:
prediction_summary_dir = source_dir_path + dir + '/0/'
# get prediction summary file
files_list = FileList.ls(path=prediction_summary_dir,
recursion=False,
mode=settings.FLS_FILE_ONLY)
for file in files_list:
print(file.full)
#print(file.version())
# read summary file and get detail information
# tissue#, label_type, label_name
summary_list = CSVLoader.csv2list(file_name=file.full)
# add overall rows
for row in summary_list[1:]:
if row[3] == 'overall':
tissue = File.tissue_info(row[2])
label_type = File.label_type(row[2])
label_name = File.label_name(row[2])
exp_num = version_info.build
# set header with additional columns (seq_type, label_type, label_name, tissue#)
if aggregated_summary is None:
header = summary_list[0]
if seq_type == "Reduced Alphabet":
header[0:1] = ['Exp_Num', 'Mapping ID']
header[3:1] = [
'seq_type', 'label_type', 'label_name',
'tissue_num'
]
else:
header[0] = 'Exp_Num'
header[2:1] = [
'seq_type', 'label_type', 'label_name',
'tissue_num'
]
aggregated_summary = list()
aggregated_summary.append(header)
data = row
if seq_type == "Reduced Alphabet":
data[0:1] = [exp_num, data[0]]
data[3:1] = [
seq_type, label_type, label_name, tissue
]
else:
data[0] = exp_num
data[2:1] = [
seq_type, label_type, label_name, tissue
]
aggregated_summary.append(data)
# write aggregated summary
if seq_type == "Reduced Alphabet":
file_name = os.path.join(
source_dir_path + '../1_summary/',
'prediction_summary_aggregated_RA_' + settings.DEV_VERSION +
'.csv')
else:
file_name = os.path.join(
source_dir_path + '../1_summary/',
'prediction_summary_aggregated_' + main_exp_num + '_' +
settings.DEV_VERSION + '.csv')
CSVLoader.list2csv(list_data=aggregated_summary, filename=file_name)
print('{} has been created.'.format(file_name))
def set_version(self, version):
self.version = VersionManager(version)
def get_feature_dirs(loc):
dirs = os.listdir(loc)
feature_dirs = [dir for dir in dirs if VersionManager.validate_pattern(dir)]
return feature_dirs