Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update data_processing.py #4

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Update data_processing.py #4

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
a854293
Update data_processing.py
KemalOrer Feb 6, 2025
dd6b254
Update data_processing.py
KemalOrer Feb 6, 2025
626c779
Update data_processing.py
KemalOrer Mar 3, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
78 changes: 64 additions & 14 deletions data_processing.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,14 +36,20 @@ def get_chemblid_smiles_inchi_dict(smiles_inchi_fl):


def save_comp_imgs_from_smiles(tar_id, comp_id, smiles, rotations, target_prediction_dataset_path, SIZE=300, rot_size=300):

mol = Chem.MolFromSmiles(smiles)

if mol is None:
print(f"Invalid SMILES: {smiles}")
return
"""
Draw.DrawingOptions.atomLabelFontSize = 55

Draw.DrawingOptions.dotsPerAngstrom = 100

Draw.DrawingOptions.bondLineWidth = 1.5

"""

base_path = os.path.join(target_prediction_dataset_path, tar_id, "imgs")

if not os.path.exists(base_path):
Expand Down Expand Up @@ -88,15 +94,20 @@ def process_smiles(smiles_data):
rotations = [(0, "_0"), *[(angle, f"_{angle}") for angle in range(10, 360, 10)]]
local_dict = {test_val_train_situation: []}
try:

save_comp_imgs_from_smiles(targetid, compound_id, current_smiles, rotations, target_prediction_dataset_path)

for i in range(0,360,10):
local_dict[test_val_train_situation].append([compound_id + "_" + str(i), int(act_inact)])
if os.path.exists(os.path.join(target_prediction_dataset_path, targetid, "imgs","{}_0.png".format(compound_id))):

for i in range(0,360,10):
local_dict[test_val_train_situation].append([compound_id + "_" + str(i), int(act_inact)])
#print(local_dict[test_val_train_situation].append([compound_id + "_" + str(i), int(act_inact)]))
else:
print(compound_id," cannot create image")
except:

print(compound_id , targetid)
pass

return local_dict

def generate_images(smiles_file, targetid, max_cores,tar_train_val_test_dict,target_prediction_dataset_path):
Expand All @@ -108,12 +119,14 @@ def generate_images(smiles_file, targetid, max_cores,tar_train_val_test_dict,tar
test_val_train_situations = pd.read_csv(smiles_file)["test_val_train"].tolist()
smiles_data_list = [(smiles, compound_ids[i], target_prediction_dataset_path, targetid,act_inact_situations[i],test_val_train_situations[i]) for i, smiles in enumerate(smiles_list)]


start_time = time.time()

with ProcessPoolExecutor(max_workers=max_cores) as executor:
results = list(executor.map(process_smiles, smiles_data_list))
end_time = time.time()

print("result" , len(results))
for result in results:
for key, value in result.items():
tar_train_val_test_dict[key].extend(value)
Expand Down Expand Up @@ -167,7 +180,7 @@ def get_act_inact_list_for_all_targets(fl):
act_inact_dict[chembl_target_id][1] = inact_list
return act_inact_dict

def create_act_inact_files_for_targets(fl, target_id, chembl_version, pchembl_threshold=6, target_prediction_dataset_path=None):
def create_act_inact_files_for_targets(fl, target_id, chembl_version, pchembl_threshold,scaffold, target_prediction_dataset_path=None):
# Create target directory if it doesn't exist
target_dir = os.path.join(target_prediction_dataset_path, target_id)
os.makedirs(target_dir, exist_ok=True)
Expand All @@ -179,8 +192,7 @@ def create_act_inact_files_for_targets(fl, target_id, chembl_version, pchembl_th
pre_filt_chembl_df['activity_label'] = (pre_filt_chembl_df['pchembl_value'] >= pchembl_threshold).astype(int)

# Now split the labeled data
train_ids, val_ids, test_ids = train_val_test_split(pre_filt_chembl_df, split_ratios=(0.8, 0.1, 0.1),
scaffold_split=True)
train_ids, val_ids, test_ids = train_val_test_split(pre_filt_chembl_df,scaffold ,split_ratios=(0.8, 0.1, 0.1))

# Create separate dataframes for train/val/test
train_df = pre_filt_chembl_df[pre_filt_chembl_df['molecule_chembl_id'].isin(train_ids)]
Expand Down Expand Up @@ -294,13 +306,48 @@ def create_act_inact_files_similarity_based_neg_enrichment_threshold(act_inact_f
act_inact_count_fl.close()
act_inact_comp_fl.close()

def create_final_randomized_training_val_test_sets(activity_data,max_cores,targetid,target_prediction_dataset_path, pchembl_threshold=6):
def create_final_randomized_training_val_test_sets(activity_data,max_cores,scaffold,targetid,target_prediction_dataset_path,moleculenet ,pchembl_threshold):



if(moleculenet):

pandas_df = pd.read_csv(activity_data)

chemblid_smiles_dict = get_chemblid_smiles_inchi_dict(activity_data)
pandas_df = pandas_df.head(200) #HERE , you can run the code for preview


pandas_df.rename(columns={pandas_df.columns[0]: "canonical_smiles", pandas_df.columns[-1]: "target"}, inplace=True)
pandas_df = pandas_df[["canonical_smiles", "target"]]

#pandas_df["molecule_chembl_id"] = [f"HIV{i+1}" for i in range(len(pandas_df))]

pandas_df["molecule_chembl_id"] = [f"{targetid}{i+1}" for i in range(len(pandas_df))]

act_ids = pandas_df[pandas_df["target"] == 1]["molecule_chembl_id"].tolist()
inact_ids = pandas_df[pandas_df["target"] == 0]["molecule_chembl_id"].tolist()
act_inact_dict = {targetid: [act_ids, inact_ids]}


moleculenet_dict = {}
for i, row_ in pandas_df.iterrows():
cid = row_["molecule_chembl_id"]
smi = row_["canonical_smiles"]
moleculenet_dict[cid] = ["dummy1", "dummy2", "dummy3", smi]
chemblid_smiles_dict = moleculenet_dict


else:

chemblid_smiles_dict = get_chemblid_smiles_inchi_dict(activity_data)

create_act_inact_files_for_targets(activity_data, targetid, "chembl", pchembl_threshold, target_prediction_dataset_path)
create_act_inact_files_for_targets(activity_data, targetid, "chembl", pchembl_threshold,scaffold, target_prediction_dataset_path)

act_inact_dict = get_act_inact_list_for_all_targets("{}/{}/{}_preprocessed_filtered_act_inact_comps_pchembl_{}.tsv".format(target_prediction_dataset_path, targetid, "chembl", pchembl_threshold))

#print(act_inact_dict)

act_inact_dict = get_act_inact_list_for_all_targets("{}/{}/{}_preprocessed_filtered_act_inact_comps_pchembl_{}.tsv".format(target_prediction_dataset_path, targetid, "chembl", pchembl_threshold))
print(len(act_inact_dict))

for tar in act_inact_dict:

Expand Down Expand Up @@ -391,6 +438,7 @@ def create_final_randomized_training_val_test_sets(activity_data,max_cores,targe

smiles_file = last_smiles_file

print("len smiles file" , len(smiles_file))
initialize_dirs(targetid , target_prediction_dataset_path)
generate_images(smiles_file , targetid , max_cores , tar_train_val_test_dict,target_prediction_dataset_path)

Expand All @@ -401,7 +449,7 @@ def create_final_randomized_training_val_test_sets(activity_data,max_cores,targe
with open(os.path.join(target_prediction_dataset_path, tar, 'train_val_test_dict.json'), 'w') as fp:
json.dump(tar_train_val_test_dict, fp)

def train_val_test_split(smiles_file, split_ratios=(0.8, 0.1, 0.1), scaffold_split=True):
def train_val_test_split(smiles_file, scaffold_split,split_ratios=(0.8, 0.1, 0.1)):
"""
Split data into train/val/test sets using either random or scaffold-based splitting

Expand Down Expand Up @@ -479,8 +527,10 @@ def __getitem__(self, index):
img_paths = [os.path.join(self.training_dataset_path, "imgs", "{}.png".format(comp_id))]


img_path = random.choice([path for path in img_paths if os.path.exists(path)])

img_path = random.choice([path for path in img_paths if os.path.exists(path)])


if not os.path.exists(img_path):
raise FileNotFoundError(f"Image not found for compound ID: {comp_id}")

Expand Down