def collectSvcData(self, dc="high"):
"""
construct the data chart for SVC,
only the -high-DecoyRepEner.txt and -low-DecoyRepEner.txt used
"""
if (dc == "high"):
raw_data_paths = suffix.getAbsPaths(self.raw_data_dir, self.extds[0]) # high decoys
elif (dc == "low"):
raw_data_paths = suffix.getAbsPaths(self.raw_data_dir, self.extds[1]) # low decoys
elif (dc == "all"):
raw_data_paths = suffix.getAbsPaths(self.raw_data_dir, self.extds[2]) # all decoys
else:
print "dc set to be high or low"
suffix.checkPaths(raw_data_paths)
tmp_data = []
for fn in raw_data_paths:
tmp_data.append(pd.read_csv(fn))
self.svc_sheet = pd.concat(tmp_data, ignore_index=True)
self.svc_sheet.headers = self.rep_ener_columns
self.svc_sheet = self.svc_sheet.drop(['Unnamed: 0'], axis=1) # rule out trash columns
self.svc_sheet = self.svc_sheet.sort(columns='mcc', ascending=False)
self.svc_sheet.index = range(len(self.svc_sheet.index))
def main(rootdir, extd, output_file = 'svm_in'):
"""
read multiple xxx-DecoyRepEner.txt files, concat them
convert to the format that libsvm accept
"""
files_to_read = suffix.getAbsPaths(rootdir, extd)
pprint.pprint(files_to_read)
data = getConcatedData(files_to_read)
data.to_csv(output_file, sep=' ',index=False, header=False)
def main(rootdir, extd, output_file = 'svm_in'):
"""
read multiple xxx-DecoyRepEner.txt files, concat them
convert to the format that libsvm accept
"""
files_to_read = suffix.getAbsPaths(rootdir, extd)
data = Concat(files_to_read)
data = reFormat(data)
data.to_csv(output_file, sep=' ',index=False, header=False)
def __init__(self, ff='-0.4.ff', dc='high', svc='svc', svr='svr', data_base_dir='/work/jaydy/dat/output/output/FF_opt/', devs=3):
self.ff = ff
self.raw_data_dir = data_base_dir + ff[1:] # raw input data dir
self.dc = dc # high, low, all
self.extd = '-' + dc + '-DecoyRepEner.txt' # extend name
self.extds = ['-high-DecoyRepEner.txt', '-low-DecoyRepEner.txt', '-all-DecoyRepEner.txt']
self.raw_data_paths = suffix.getAbsPaths(self.raw_data_dir, self.extd) # paths of the raw input data files
self.concated_fn = dc + '-raw' # name of the concated file
self.rep_ener_columns = ['_evdw', '_eele', '_epmf', '_ehpc', '_ehdb', '_edst', '_epsp', '_ekde', '_elhm', 'mcc']
self.svc = svc # name
self.svr = svr # name
self.svc_sheet = pd.DataFrame()
self.svr_sheet = pd.DataFrame()
self.devs = devs