File size: 12,369 Bytes
ed339ed |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 |
import os
import sys
sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
import protac_degradation_predictor as pdp
from collections import defaultdict
import warnings
import logging
from typing import Literal
from sklearn.preprocessing import OrdinalEncoder
from tqdm import tqdm
import pandas as pd
import numpy as np
import pytorch_lightning as pl
from rdkit import DataStructs
root = logging.getLogger()
root.setLevel(logging.DEBUG)
handler = logging.StreamHandler(sys.stdout)
handler.setLevel(logging.DEBUG)
formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
handler.setFormatter(formatter)
root.addHandler(handler)
def get_random_split_indices(active_df: pd.DataFrame, test_split: float) -> pd.Index:
""" Get the indices of the test set using a random split.
Args:
active_df (pd.DataFrame): The DataFrame containing the active PROTACs.
test_split (float): The percentage of the active PROTACs to use as the test set.
Returns:
pd.Index: The indices of the test set.
"""
test_df = active_df.sample(frac=test_split, random_state=42)
return test_df.index
def get_e3_ligase_split_indices(active_df: pd.DataFrame) -> pd.Index:
""" Get the indices of the test set using the E3 ligase split.
Args:
active_df (pd.DataFrame): The DataFrame containing the active PROTACs.
Returns:
pd.Index: The indices of the test set.
"""
encoder = OrdinalEncoder()
active_df['E3 Group'] = encoder.fit_transform(active_df[['E3 Ligase']]).astype(int)
test_df = active_df[(active_df['E3 Ligase'] != 'VHL') & (active_df['E3 Ligase'] != 'CRBN')]
return test_df.index
def get_smiles2fp_and_avg_tanimoto(protac_df: pd.DataFrame) -> tuple:
""" Get the SMILES to fingerprint dictionary and the average Tanimoto similarity.
Args:
protac_df (pd.DataFrame): The DataFrame containing the PROTACs.
Returns:
tuple: The SMILES to fingerprint dictionary and the average Tanimoto similarity.
"""
unique_smiles = protac_df['Smiles'].unique().tolist()
smiles2fp = {}
for smiles in tqdm(unique_smiles, desc='Precomputing fingerprints'):
smiles2fp[smiles] = pdp.get_fingerprint(smiles)
# # Get the pair-wise tanimoto similarity between the PROTAC fingerprints
# tanimoto_matrix = defaultdict(list)
# for i, smiles1 in enumerate(tqdm(protac_df['Smiles'].unique(), desc='Computing Tanimoto similarity')):
# fp1 = smiles2fp[smiles1]
# # TODO: Use BulkTanimotoSimilarity for better performance
# for j, smiles2 in enumerate(protac_df['Smiles'].unique()[i:]):
# fp2 = smiles2fp[smiles2]
# tanimoto_dist = 1 - DataStructs.TanimotoSimilarity(fp1, fp2)
# tanimoto_matrix[smiles1].append(tanimoto_dist)
# avg_tanimoto = {k: np.mean(v) for k, v in tanimoto_matrix.items()}
# protac_df['Avg Tanimoto'] = protac_df['Smiles'].map(avg_tanimoto)
tanimoto_matrix = defaultdict(list)
fps = list(smiles2fp.values())
# Compute all-against-all Tanimoto similarity using BulkTanimotoSimilarity
for i, (smiles1, fp1) in enumerate(tqdm(zip(unique_smiles, fps), desc='Computing Tanimoto similarity', total=len(fps))):
similarities = DataStructs.BulkTanimotoSimilarity(fp1, fps[i:]) # Only compute for i to end, avoiding duplicates
for j, similarity in enumerate(similarities):
distance = 1 - similarity
tanimoto_matrix[smiles1].append(distance) # Store as distance
if i != i + j:
tanimoto_matrix[unique_smiles[i + j]].append(distance) # Symmetric filling
# Calculate average Tanimoto distance for each unique SMILES
avg_tanimoto = {k: np.mean(v) for k, v in tanimoto_matrix.items()}
protac_df['Avg Tanimoto'] = protac_df['Smiles'].map(avg_tanimoto)
smiles2fp = {s: np.array(fp) for s, fp in smiles2fp.items()}
return smiles2fp, protac_df
def get_tanimoto_split_indices(
active_df: pd.DataFrame,
active_col: str,
test_split: float,
n_bins_tanimoto: int = 200,
) -> pd.Index:
""" Get the indices of the test set using the Tanimoto-based split.
Args:
active_df (pd.DataFrame): The DataFrame containing the active PROTACs.
n_bins_tanimoto (int): The number of bins to use for the Tanimoto similarity.
Returns:
pd.Index: The indices of the test set.
"""
tanimoto_groups = pd.cut(active_df['Avg Tanimoto'], bins=n_bins_tanimoto).copy()
encoder = OrdinalEncoder()
active_df['Tanimoto Group'] = encoder.fit_transform(tanimoto_groups.values.reshape(-1, 1)).astype(int)
# Sort the groups so that samples with the highest tanimoto similarity,
# i.e., the "less similar" ones, are placed in the test set first
tanimoto_groups = active_df.groupby('Tanimoto Group')['Avg Tanimoto'].mean().sort_values(ascending=False).index
test_df = []
# For each group, get the number of active and inactive entries. Then, add those
# entries to the test_df if: 1) the test_df lenght + the group entries is less
# 20% of the active_df lenght, and 2) the percentage of True and False entries
# in the active_col in test_df is roughly 50%.
for group in tanimoto_groups:
group_df = active_df[active_df['Tanimoto Group'] == group]
if test_df == []:
test_df.append(group_df)
continue
num_entries = len(group_df)
num_active_group = group_df[active_col].sum()
num_inactive_group = num_entries - num_active_group
tmp_test_df = pd.concat(test_df)
num_entries_test = len(tmp_test_df)
num_active_test = tmp_test_df[active_col].sum()
num_inactive_test = num_entries_test - num_active_test
# Check if the group entries can be added to the test_df
if num_entries_test + num_entries < test_split * len(active_df):
# Add anything at the beggining
if num_entries_test + num_entries < test_split / 2 * len(active_df):
test_df.append(group_df)
continue
# Be more selective and make sure that the percentage of active and
# inactive is balanced
if (num_active_group + num_active_test) / (num_entries_test + num_entries) < 0.6:
if (num_inactive_group + num_inactive_test) / (num_entries_test + num_entries) < 0.6:
test_df.append(group_df)
test_df = pd.concat(test_df)
return test_df.index
def get_target_split_indices(active_df: pd.DataFrame, active_col: str, test_split: float) -> pd.Index:
""" Get the indices of the test set using the target-based split.
Args:
active_df (pd.DataFrame): The DataFrame containing the active PROTACs.
active_col (str): The column containing the active/inactive information.
test_split (float): The percentage of the active PROTACs to use as the test set.
Returns:
pd.Index: The indices of the test set.
"""
encoder = OrdinalEncoder()
active_df['Uniprot Group'] = encoder.fit_transform(active_df[['Uniprot']]).astype(int)
test_df = []
# For each group, get the number of active and inactive entries. Then, add those
# entries to the test_df if: 1) the test_df lenght + the group entries is less
# 20% of the active_df lenght, and 2) the percentage of True and False entries
# in the active_col in test_df is roughly 50%.
# Start the loop from the groups containing the smallest number of entries.
for group in reversed(active_df['Uniprot'].value_counts().index):
group_df = active_df[active_df['Uniprot'] == group]
if test_df == []:
test_df.append(group_df)
continue
num_entries = len(group_df)
num_active_group = group_df[active_col].sum()
num_inactive_group = num_entries - num_active_group
tmp_test_df = pd.concat(test_df)
num_entries_test = len(tmp_test_df)
num_active_test = tmp_test_df[active_col].sum()
num_inactive_test = num_entries_test - num_active_test
# Check if the group entries can be added to the test_df
if num_entries_test + num_entries < test_split * len(active_df):
# Add anything at the beggining
if num_entries_test + num_entries < test_split / 2 * len(active_df):
test_df.append(group_df)
continue
# Be more selective and make sure that the percentage of active and
# inactive is balanced
if (num_active_group + num_active_test) / (num_entries_test + num_entries) < 0.6:
if (num_inactive_group + num_inactive_test) / (num_entries_test + num_entries) < 0.6:
test_df.append(group_df)
test_df = pd.concat(test_df)
return test_df.index
def main(
active_col: str = 'Active (Dmax 0.6, pDC50 6.0)',
test_split: float = 0.1,
studies: str | Literal['all', 'standard', 'e3_ligase', 'similarity', 'target'] = 'all',
):
""" Get and save the datasets for the different studies.
Args:
active_col (str): The column containing the active/inactive information. It should be in the format 'Active (Dmax N, pDC50 M)', where N and M are the thresholds float values for Dmax and pDC50, respectively.
test_split (float): The percentage of the active PROTACs to use as the test set.
studies (str): The type of studies to save dataset for. Options: 'all', 'standard', 'e3_ligase', 'similarity', 'target'.
"""
pl.seed_everything(42)
# Set the Column to Predict
active_name = active_col.replace(' ', '_').replace('(', '').replace(')', '').replace(',', '')
# Get Dmax_threshold from the active_col
Dmax_threshold = float(active_col.split('Dmax')[1].split(',')[0].strip('(').strip(')').strip())
pDC50_threshold = float(active_col.split('pDC50')[1].strip('(').strip(')').strip())
# Load the PROTAC dataset
protac_df = pd.read_csv('../data/PROTAC-Degradation-DB.csv')
# Map E3 Ligase Iap to IAP
protac_df['E3 Ligase'] = protac_df['E3 Ligase'].str.replace('Iap', 'IAP')
protac_df[active_col] = protac_df.apply(
lambda x: pdp.is_active(x['DC50 (nM)'], x['Dmax (%)'], pDC50_threshold=pDC50_threshold, Dmax_threshold=Dmax_threshold), axis=1
)
_, protac_df = get_smiles2fp_and_avg_tanimoto(protac_df)
## Get the test sets
test_indeces = {}
active_df = protac_df[protac_df[active_col].notna()].copy()
# Remove legacy column 'Active - OR' if it exists
if 'Active - OR' in active_df.columns:
active_df.drop(columns='Active - OR', inplace=True)
if studies == 'standard' or studies == 'all':
test_indeces['standard'] = get_random_split_indices(active_df, test_split)
if studies == 'target' or studies == 'all':
test_indeces['target'] = get_target_split_indices(active_df, active_col, test_split)
if studies == 'e3_ligase' or studies == 'all':
test_indeces['e3_ligase'] = get_e3_ligase_split_indices(active_df)
if studies == 'similarity' or studies == 'all':
test_indeces['similarity'] = get_tanimoto_split_indices(active_df, active_col, test_split)
# Make directory for studies datasets if it does not exist
data_dir = '../data/studies'
if not os.path.exists(data_dir):
os.makedirs(data_dir)
# Cross-Validation Training
for split_type, indeces in test_indeces.items():
test_df = active_df.loc[indeces].copy()
train_val_df = active_df[~active_df.index.isin(test_df.index)].copy()
# Save the datasets
train_val_perc = f'{int((1 - test_split) * 100)}'
test_perc = f'{int(test_split * 100)}'
train_val_filename = f'{data_dir}/{split_type}_train_val_{train_val_perc}split_{active_name}.csv'
test_filename = f'{data_dir}/{split_type}_test_{test_perc}split_{active_name}.csv'
print('')
print(f'Saving train_val datasets as: {train_val_filename}')
print(f'Saving test datasets as: {test_filename}')
train_val_df.to_csv(train_val_filename, index=False)
test_df.to_csv(test_filename, index=False)
if __name__ == '__main__':
main() |