def fetch_fastas_for_DUDE():
if not os.path.exists(
os.path.join(BLAST_MAIN_FOLDER, 'fastas_from_dude.txt')):
log(f'Starting fetching FASTA\'s for DUDE')
dude_uni_ids = DUDE_uniID_from_folder(DUDE_PATH)
with open(os.path.join(BLAST_MAIN_FOLDER, 'fastas_from_dude.txt'),
'w') as handle:
for uniprot_id in dude_uni_ids:
pdb_ids = get_pdbs_from_unicode(uniprot_id)
pdbs_dict = dict()
for pdb in pdb_ids:
try:
if check_if_pdb_xray(pdb):
pdb_fasta = get_pdb_fasta(pdb)
pdbs_dict[uniprot_id + '-' +
pdb] = pdb_fasta.strip('"')
except HTTPError:
log(f'Failed fetching FASTA sequence for: {pdb}')
pdbs_dict2 = dict()
for key, value in pdbs_dict.items():
if value not in list(pdbs_dict2.values()):
pdbs_dict2[key] = value
for key, value in pdbs_dict2.items():
handle.write(f'>{key}\n{value}\n')
else:
log('DUD-E fastas already exists.')
def fetch_fastas_for_DEKOIS():
if not os.path.exists(
os.path.join(BLAST_MAIN_FOLDER, 'fastas_from_dekois.txt')):
log('Starting fetching FASTA\' s for DEKOIS')
dekois_ligands, _ = DEKOIS_uniID_from_folder(DEKOIS_PATH)
dekois_ligands_uni_ids = set()
for i in dekois_ligands:
try:
for k in dekois_ligands[i]['Uniprot_ID'].split():
dekois_ligands_uni_ids.add(k)
except KeyError:
log(f'No Uniprot ID for {i}')
with open(os.path.join(BLAST_MAIN_FOLDER, 'fastas_from_dekois.txt'),
'w') as handle:
for uniprot_id in dekois_ligands_uni_ids:
pdb_ids = get_pdbs_from_unicode(uniprot_id)
pdbs_dict = dict()
for pdb in pdb_ids:
try:
if check_if_pdb_xray(pdb):
pdb_fasta = get_pdb_fasta(pdb)
pdbs_dict[uniprot_id + '-' +
pdb] = pdb_fasta.strip('"')
except HTTPError:
log(f'Failed fetching FASTA sequence for: {pdb}')
pdbs_dict2 = dict()
for key, value in pdbs_dict.items():
if value not in list(pdbs_dict2.values()):
pdbs_dict2[key] = value
for key, value in pdbs_dict2.items():
handle.write(f'>{key}\n{value}\n')
else:
log('DEKOIS fastas already exists.')
def fetch_fastas_for_chembl(csv_path, output):
if not os.path.exists(output):
main_table = pd.read_csv(csv_path, index_col=0)
log(f'Downloading fastas for {len(main_table)} targets')
with open(output, 'w') as handle:
for index, row in main_table.iterrows():
chembl_name = row['ChEMBL ID']
pdb_name = row['main_PDB_structure']
fasta = get_pdb_fasta(row['main_PDB_structure'])
handle.write(f'>{chembl_name}-{pdb_name}\n{fasta}\n')
else:
log(f'{output} already exists!')
def choose_primary_pdb_for_chembl(csv_path):
main_table = pd.read_csv(csv_path, index_col=0)
main_table['main_PDB_structure'] = ''
for index, row in main_table.iterrows():
pdbs = row['PDB_entry'].split()
fasta = [get_pdb_fasta(i) for i in pdbs]
best = ''
for seq in fasta:
if len(seq) > len(best):
best = seq
if best != '':
best_pdb = pdbs[fasta.index(best)]
main_table.at[index, 'main_PDB_structure'] = best_pdb
main_table.to_csv(csv_path)
def choose_primary_pdb_for_chembl(main_table):
main_table['main_PDB_structure'] = ''
for index, row in main_table.iterrows():
pdbs = row['PDB_entry'].split()
log(f'Fetching PDBs for {index}, {len(pdbs)} PDBs to fetch.')
fasta = [get_pdb_fasta(i) for i in pdbs]
best = ''
for seq in fasta:
if len(seq) > len(best):
best = seq
if best != '':
best_pdb = pdbs[fasta.index(best)]
main_table.at[index, 'main_PDB_structure'] = best_pdb
main_table = main_table.sort_values(
'Active_compounds',
ascending=False).drop_duplicates(subset='main_PDB_structure')
main_table = main_table[main_table['main_PDB_structure'] != '']
return main_table
def make_blast_csv(master_path, blasts=None):
main_table = pd.read_csv(master_path, index_col=0)
output_df = main_table.loc[:, [
'ChEMBL ID', 'main_PDB_structure', 'Active_compounds',
'Inactive_compounds'
]]
output_file = os.path.join(BLAST_MAIN_FOLDER, 'chembl_blast_results.csv')
for db in blasts:
#chembl_Id, hit_ID, identity%, evalue
blast_results = load_blast_results(db, [0, 1, 2, 6, 7, 10])
db_name = str(os.path.split(db)[-1]).split('_')[0].split('-')[1]
output_df[f'identity%_{db_name}'] = ''
output_df[f'evalue_{db_name}'] = ''
output_df[f'target_name_{db_name}'] = ''
output_df[f'query_alignment_length_{db_name}'] = ''
output_df[f'total_query_length_{db_name}'] = ''
output_df[f'alignment_to_total_ratio_{db_name}'] = ''
name = ''
for chembl_id, db_target, identity, q_start, q_end, evalue in blast_results:
if name not in chembl_id or name == '':
query_pdb_smile_len = len(
get_pdb_fasta(chembl_id.split('-')[-1]))
query_alignment_length = q_end - q_start
name = chembl_id.split('-')[0]
output_df.at[name, f'identity%_{db_name}'] = identity
output_df.at[name, f'evalue_{db_name}'] = evalue
output_df.at[name, f'target_name_{db_name}'] = db_target
output_df.at[
name,
f'query_alignment_length_{db_name}'] = query_alignment_length
output_df.at[
name,
f'total_query_length_{db_name}'] = query_pdb_smile_len
output_df.at[
name,
f'alignment_to_total_ratio_{db_name}'] = query_alignment_length / query_pdb_smile_len
output_df.to_csv(output_file)