def prepare_dataset(sdffile, dest=None, rename=True, conformations=False, overwrite=False):
""" This method bootstraps the analysis of Ames data.
- Rename the compounds
- Merge train/test
- Generate 3D conformations
- Save "master" and "saliviewer" tables
- Redirects stdout/stderr to a "prepare.log" file
"""
root, name = op.split(sdffile)
name = op.splitext(name)[0]
if not dest: dest = root
dest_sdf = op.join(dest, name + '-prepared.sdf')
master_table = op.join(dest, name + '-prepared-master.csv')
sali_table = op.join(dest, name + '-prepared-saliviewer.csv')
if op.exists(dest_sdf) and not overwrite:
print '%s is already there and not overwriting requested' % dest_sdf
else:
print 'Reading %s' % sdffile
mols = list(pybel.readfile('sdf', sdffile))
print '\tCreating dataset root: %s' % dest
if not op.exists(dest):
os.makedirs(dest)
if rename:
print '\tRenaming the compounds to keep track of the provenance'
rename_mols_by_index(mols, name + '-')
if conformations:
print '\tGenerating conformations'
for mol in mols:
if not any(name in mol.title for name in ('train-3988', 'train-4205', 'dsstox-4205', 'dsstox-4206')):
try:
print 'Conformation for %s' % mol.title
mol.make3D()
except Exception:
print 'Error computing a 3D conformation for %s' % mol.title
print '\tSaving compounds'
save_mols(mols, dest_sdf)
print '\tCreating \"master\" table: %s' % master_table
create_master_table(dest_sdf, master_table, fields=['Activity'])
print '\tCreating \"saliviewer\" table: %s' % sali_table
create_saliviewer_input(master_table, sali_table)
return dest_sdf, master_table
def aid2sdf(sdf, csv, dest=None):
""" Reads a pubchem bioassay results and merge it with the SDF file """
#Read the known activities to a dictionary
activities = {}
for activity in open(csv).readlines()[1:]:
value = activity.split(',')[5]
molid = activity.split(',')[2]
activities[molid] = value
#Save the activity to each molecule
mols = list(pybel.readfile('sdf', sdf))
for mol in mols:
activity = activities[mol.title]
if activity == 'Active': actual_activity = '1'
elif activity == 'Inactive': actual_activity = '0'
else: actual_activity = 'Missing'
mol.data['Activity'] = actual_activity
if dest:
save_mols(mols, dest)
return mols
cas = data[0].data['CAS_NO']
if not cas in can_dupes2:
can_dupes2[cas] = [data]
else:
can_dupes2[cas].append(data)
for cas in can_dupes2.keys():
groups = can_dupes2[cas]
if len(groups) > 1:
print 'compound with cas=%s is considered different by OB canonical smiles'%cas
for group in groups:
print group[0].write('can').strip()
print '-'*80
union = sorted([dupe[0] for dupe in cas_dupes.values()], key=lambda mol: mol.title)
save_mols(union, op.join(root, 'mutagenicity-all-cas-union.sdf'))
print '\t\tUnion size=%d' % len(union)
dest_sdf = op.join(root, 'mutagenicity-all-cas-union-prepared.sdf')
prepare_dataset(op.join(root, 'mutagenicity-all-cas-union.sdf'), rename=False, conformations=True)
#Depict the molecules
depict(dest_sdf)
#Molecular descriptors
print 'Computing fingerprints via JCompoundMapper' #TODO: Extract-method this
jcm_fingerprint(dest_sdf, ('ECFP', 'ECFPVariant', 'PHAP3POINT2D', 'SHED', 'DFS', 'RAD2D'))
jcm_fingerprint(dest_sdf, ('LSTAR', 'RAD3D', 'PHAP3POINT3D'))
print 'Computing descriptors via CDKDescUI'
cdkdescuiprops(dest_sdf, desc_types=('constitutional',))
print 'Computing spectrophores'