def cdk_desc_to_arff(directory, master_file, desc_csv, to_predict):
y = read_y_from_master(op.join(directory, master_file))
classes = infer_classes(y)
descs = op.join(directory, desc_csv)
f = open(descs, "r")
data = []
line1 = f.readline()
feature_names = ["ID"] + [name for name in line1.split()[1:]]
for line in f:
if len(line.strip()):
data.append(map(floatOrNaN, line.strip().split("\t")))
f.close()
x = np.array(data)
arff_file = op.join(directory, op.splitext(desc_csv)[0] + ".arff")
mlio.save_arff(x, y, arff_file, relation_name=to_predict, feature_names=feature_names, classes=classes)
def spectrophores_to_arff(directory, master_file, spec_csv, to_predict):
y = read_y_from_master(op.join(directory, master_file))
classes = infer_classes(y)
specs = op.join(directory, spec_csv)
f = open(specs, "r")
data = []
for line in f:
# data.append(map(lambda a: float(a.strip()), line.split(',')))
if len(line.strip()):
data.append(map(floatOrNaN, line.strip().split(",")[:-1]))
x = np.array(data)
feature_names = ["ID"]
f.close()
for i in range(48):
feature_names.append("Spec" + str(i))
arff_file = op.join(directory, op.splitext(spec_csv)[0] + ".arff")
mlio.save_arff(x, y, arff_file, relation_name=to_predict, feature_names=feature_names, classes=classes)
def cdk_fpt_to_arff(directory, master_file, fpt_csv, to_predict, fpt_type):
y = read_y_from_master(op.join(directory, master_file))
classes = infer_classes(y)
fpts = op.join(directory, fpt_csv)
f = open(fpts, "r")
x, _ = cdkdeskuifps2dense(fpts, keep_id=True)
feature_names = ["ID"]
if fpt_type == "maccs":
for i in range(166):
feature_names.append("maccs" + str(i))
elif fpt_type == "estate":
for i in range(79):
feature_names.append("estate" + str(i))
elif fpt_type == "extended":
for i in range(1024):
feature_names.append("extended" + str(i))
else:
print "Fingerprint type currently not supported."
arff_file = op.join(directory, op.splitext(fpt_csv)[0] + ".arff")
mlio.save_arff(x, y, arff_file, relation_name=to_predict, feature_names=feature_names, classes=classes)
def prop4da(dataset):
root, name = op.split(dataset)
name = op.splitext(name)[0]
masterfile = op.join(root, name + "-master.csv")
y = read_y_from_master(masterfile)
classes = infer_classes(y)
# Process CDK descriptors
for descs in glob.glob(op.join(root, "*-cdk-*.csv")):
with open(descs) as reader:
header = reader.next()
if header.startswith("Title"):
x, features = cdkdeskui2dense(descs)
mlio.save_arff(x, y, op.splitext(descs)[0] + ".arff", feature_names=features, classes=classes)
mlio.save_tab(x, y, op.splitext(descs)[0] + ".txt", classes=classes)
else:
x, relation_name = cdkdeskuifps2dense(descs)
features = mlio.generate_names(x.shape[0], descs)
mlio.save_arff(
x,
y,
op.splitext(descs)[0] + ".arff",
relation_name=relation_name,
classes=classes,
feature_names=features,
)
mlio.save_tab(x, y, op.splitext(descs)[0] + ".txt", classes=classes)
# Process ob spectrophores
specs = op.join(root, name + "-ob-spectrophores.csv")
with open(specs) as reader:
specs = []
for line in reader:
specs.append(map(lambda a: float(a.strip()), line.split(",")))
x = np.array(specs)
feature_names = mlio.generate_names(len(specs[0]))
mlio.save_arff(x, y, op.join(root, name + "-ob-spectrophores.arff"), classes=classes, feature_names=feature_names)
mlio.save_tab(x, y, op.join(root, name + "-ob-spectrophores.txt"), classes=classes)
datasets = sorted([name for name in os.listdir(root) if op.isdir(op.join(root, name))])
for dataset in datasets:
print dataset
y = read_y(root, dataset)
classes = infer_classes(y)
#Process ob spectrophores
specs = op.join(root, dataset, dataset + '-ob-spectrophores.csv')
with open(specs) as reader:
specs = []
for line in reader:
specs.append(map(lambda a: float(a.strip()), line.split(',')))
x = np.array(specs)
mlio.save_arff(x, y, op.join(root, dataset, dataset + '-ob-spectrophores.arff'), classes=classes)
mlio.save_tab(x, y, op.join(root, dataset, dataset + '-ob-spectrophores.txt'), classes=classes)
#Process CDK descriptors
for descs in glob.glob(op.join(root, dataset, '*-cdk-*.csv')):
with open(descs) as reader:
header = reader.next()
if header.startswith('Title'):
x, features = cdkdeskui2dense(descs)
mlio.save_arff(x, y, op.splitext(descs)[0] + '.arff', feature_names=features, classes=classes)
mlio.save_tab(x, y, op.splitext(descs)[0] + '.txt', classes=classes)
else:
x, name = cdkdeskuifps2dense(descs)
mlio.save_arff(x, y, op.splitext(descs)[0] + '.arff', relation_name=name, classes=classes)
mlio.save_tab(x, y, op.splitext(descs)[0] + '.txt', classes=classes)
