theshold = 0.6
fileout = open("uniprot_lig_tanamoto.txt", 'w')
for uniprot in uniprot_to_ligands_dict.keys():
fileout = open("uniprot_lig_tanamoto_flush.txt", 'a')
if (len(uniprot_to_ligands_dict[uniprot]) > 6):
liglist = uniprot_to_ligands_dict[uniprot]
print uniprot, uniprot_to_ligands_dict[uniprot]
# put to forloops, both over the ligands (do uppper diagonal), here and cal tanamotos
for i in range(len(liglist)):
lig_i = liglist[i]
if (lig_i in lig_smiles):
smiles_i = lig_smiles[lig_i]
fp1 = lig_fp[lig_i]
else:
smiles_i = spfl.scrape_pdb_for_lig_smiles(lig_i)
fp1 = tancal.fingerprint(smiles_i)
lig_smiles[lig_i] = smiles_i
lig_fp[lig_i] = fp1
for j in range(i + 1, len(liglist)):
lig_j = liglist[j]
if (lig_j in lig_smiles):
smiles_j = lig_smiles[lig_j]
fp2 = lig_fp[lig_j]
else:
smiles_j = spfl.scrape_pdb_for_lig_smiles(lig_j)
fp2 = tancal.fingerprint(smiles_j)
lig_smiles[lig_j] = smiles_j
lig_fp[lig_j] = fp2
#smiles_j = spfl.scrape_pdb_for_lig_smiles(lig_j)
# print lig_list
for lig in lig_list:
#if lig in lig_dict: # if the ligand is already in the dictionary then dont looked up smile
# continue
#if lig in ions: ## skip if any ion
if lig in not_a_ligand: ## eg. skip if any ion
#print lig, "in", not_a_ligand
continue
##if lig in cofactors:
##if not (lig in lig_dict) and not (lig in ions) :
if not (
lig in lig_dict
): # if the ligand is already in the dictionary then dont looked up smile
smiles = spfl.scrape_pdb_for_lig_smiles(lig)
lig_dict[lig] = smiles
entry = [pdb, lig]
pdbtolig_filehandle.write("%s %s\n" % (pdb, lig))
pdb_to_ligand.append(entry)
pdbtolig_filehandle.close()
pdbtouniprot_filehandle.close()
#for lig in unprot_list:
## Write out information to files
## consider creating a ProgreSQP to store this information.
unifilehandle = open('uniprotfile.txt', 'w')
for id in uniprot_dict.keys():
