File size: 11,319 Bytes
6f1e670 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 | import pandas as pd
import numpy as np
from pathlib import Path
import json, hashlib
import re
from model.splitters import *
from model.featurizers import *
from model.predictors import *
from model.proposers import *
class TrainingCache:
"""
A cache class for storing training results with an index-based lookup system.
Attributes:
dir (Path): Directory path for cache storage
index_path (Path): Path to the index CSV file
index (pd.DataFrame): DataFrame containing cache metadata and lookup information
"""
def __init__(self, cache_dir: str | Path):
"""
Initialize the TrainingCache.
Args:
cache_dir (str | Path): Directory path where cache files will be stored
"""
self.dir = Path(cache_dir)
self.dir.mkdir(parents=True, exist_ok=True)
self.index_path = self.dir / "index.csv"
if self.index_path.exists():
# Just load it back; columns will include whatever keys you added
self.index = pd.read_csv(self.index_path)
else:
# Start empty; no need to predefine columns, they’ll be added by set()
self.index = pd.DataFrame()
def _key_id(self, keys: dict) -> str:
"""
Generate a unique hash ID for the given keys dictionary.
Args:
keys (dict): Dictionary of keys to hash
Returns:
str: MD5 hash of the serialized keys
"""
blob = json.dumps(keys, sort_keys=True, default=str)
return hashlib.md5(blob.encode()).hexdigest()
def _path(self, key_id: str) -> Path:
"""
Generate the file path for a given key ID.
Args:
key_id (str): Unique identifier for the cache entry
Returns:
Path: Path object pointing to the pickle file
"""
return self.dir / f"{key_id}.pkl"
def _check_index(self, row: pd.DataFrame) -> None:
# check if index exists
if self.index_path.exists():
self.index = pd.read_csv(self.index_path)
# overwrite existing entry in case of updating path
if not self.index.empty and "key_id" in self.index.columns:
self.index = self.index[self.index["key_id"] != row['key_id']]
self.index = pd.concat([self.index, pd.DataFrame([row])], ignore_index=True)
self.index.to_csv(self.index_path, index=False)
def get(self, keys: dict) -> pd.DataFrame | None:
"""
Check if a pkl file for run exists, if so retrieve dataframe and check index if run is in index.
Args:
keys (dict): Dictionary of keys to look up
Returns:
pd.DataFrame | None: Cached DataFrame if found, None otherwise
"""
# Check if pkl file for run exists else return None
path = self._path(self._key_id(keys))
if not path.exists():
return None
# retrieve dataframe
df = pd.read_pickle(path)
# update index
row = {
"key_id": self._key_id(keys),
"path": str(path),
"variants": df.shape[0],
**keys, # expand keys directly into columns
}
self._check_index(row)
return df
def set(self, keys: dict, df: pd.DataFrame) -> None:
"""
Store a DataFrame in the cache with the given keys and check index to update with new entry.
Args:
keys (dict): Dictionary of keys to associate with the DataFrame
df (pd.DataFrame): DataFrame to cache
"""
key_id = self._key_id(keys)
path = self._path(key_id)
# save dataframe
df.to_pickle(path)
# build row with expanded keys
row = {
"key_id": key_id,
"path": str(path),
"variants": df.shape[0],
**keys, # expand keys directly into columns
}
# update index
self._check_index(row)
def summary_df_check_dms_completion(summary_df, threshold=0.8):
def check_dms_completion(row):
fraction_dms = row['DMS_number_single_mutants'] / (row['seq_len'] * 19)
if fraction_dms >= threshold:
return fraction_dms, True
else:
return fraction_dms, False
summary_df[['fraction_dms', 'dms_threshold_met']] = summary_df.apply(check_dms_completion, axis=1, result_type='expand')
return summary_df
# function to receive dataset name, dataset filename, and sequence from dataframe row
def receive_dataset_vars(row):
# generate fasta file of sequence
dataset_name = row['DMS_id']
dataset_fname = row['DMS_filename']
sequence = row['target_seq']
return dataset_name, dataset_fname, sequence
# function to generate fasta file of sequence
def retrieve_wt_file(dataset_name, seq_dir, sequence):
output_dir = Path(seq_dir)
wt_file = output_dir / f'{dataset_name}.fasta'
if wt_file.exists():
return str(wt_file)
else:
# Prepare folder to save results
output_dir.mkdir(parents=True, exist_ok=True)
with open(wt_file, 'w') as file:
file.write(f'>{dataset_name}\n')
file.write(sequence + '\n')
return str(wt_file)
# function to preprocess dataset (mark valid multimutants), add relevant columns
def preprocess_dataset(dataset_fname, data_dir, stringency='singles'):
# options for stringency: 'singles', 'singles_or_doubles', 'singles_positions'
# check for existing processed datasets
processed_datasets_dir = os.path.join(data_dir, 'processed')
processed_datasets_all_dir = os.path.join(data_dir, 'processed', 'all')
processed_datasets_stringency_dir = os.path.join(data_dir, 'processed', stringency)
processed_all_filename = os.path.join(processed_datasets_all_dir, f'{dataset_fname}.csv')
processed_stringency_filename = os.path.join(processed_datasets_stringency_dir, f'{dataset_fname}.csv')
if os.path.exists(processed_all_filename) and os.path.exists(processed_stringency_filename):
return pd.read_csv(processed_all_filename), pd.read_csv(processed_stringency_filename)
# read csv file
working_df_head = pd.read_csv(os.path.join(data_dir, dataset_fname))
# replace colon with slash in mutant column
working_df_head['mutant'] = working_df_head['mutant'].apply(lambda x: x.replace(':', '/'))
# retrieve number of mutations
working_df_head['num_mutations'] = working_df_head['mutant'].apply(lambda x: len(x.split('/')))
# retrieve single mutants
singles = working_df_head[working_df_head['num_mutations'] == 1]['mutant'].tolist()
# get positions from single mutants
singles_positions = []
pattern = r'[A-Z]\d+[A-Z]'
for mutant in singles:
if len(re.findall(pattern, mutant)) != 0:
singles_positions.append(int(mutant[1:-1]))
singles_positions = list(set(singles_positions))
# retrieve mutations doubles
doubles = [single for double in working_df_head[working_df_head['num_mutations'] == 2]['mutant'].tolist() for single in double.split('/')]
# function to check for multimutants where are single mutants are present
def check_existing_mutants_in_singles(mutant):
mutant_list = mutant.split('/')
for m in mutant_list:
if m in singles:
pass
else:
return False
return True
def check_existing_mutants_in_singles_or_doubles(mutant):
mutant_list = mutant.split('/')
for m in mutant_list:
if m in singles or m in doubles:
pass
else:
return False
return True
def check_existing_mutants_in_singles_positions(mutant):
mutant_list = mutant.split('/')
if mutant_list[0] == 'WT':
return True
else:
for m in mutant_list:
if int(m[1:-1]) in singles_positions:
pass
else:
return False
return True
# filter dataset keeping only combo variants with all single mutants existing in the dataset
working_df_head['singles_exist'] = working_df_head['mutant'].apply(check_existing_mutants_in_singles)
working_df_head['singles_or_doubles_exist'] = working_df_head['mutant'].apply(check_existing_mutants_in_singles_or_doubles)
working_df_head['singles_positions_exist'] = working_df_head['mutant'].apply(check_existing_mutants_in_singles_positions)
if stringency == 'singles':
working_df_head_valid = working_df_head[working_df_head['singles_exist'] == True].sort_values(by='num_mutations', ascending=False)
elif stringency == 'singles_or_doubles':
working_df_head_valid = working_df_head[working_df_head['singles_or_doubles_exist'] == True].sort_values(by='num_mutations', ascending=False)
elif stringency == 'singles_positions':
working_df_head_valid = working_df_head[working_df_head['singles_positions_exist'] == True].sort_values(by='num_mutations', ascending=False)
else:
raise ValueError(f'Invalid stringency: {stringency}. Please choose from ["singles", "singles_or_doubles"].')
# save results
os.makedirs(os.path.join(processed_datasets_all_dir), exist_ok=True)
os.makedirs(os.path.join(processed_datasets_stringency_dir), exist_ok=True)
working_df_head.to_csv(processed_all_filename, index=False)
working_df_head_valid.to_csv(processed_stringency_filename, index=False)
return working_df_head, working_df_head_valid
def select_feature(encoding, protein_name, batch_size=1000):
# check if encoding_name is in the list of available encodings
if encoding not in ['esm2_15b', 'esm2_3b', 'onehot', 'georgiev', 'aaidx', 'onehot_and_esm2_15b', 'ankh_base', 'ankh_large', 'ProtT5_XL_U50_Embed']:
raise ValueError(f'Invalid encoding_name: {encoding}. Please choose from {["esm2_15b", "esm2_3b", "onehot", "georgiev", "aaidx", "ankh_base", "ankh_large", "ProtT5_XL_U50_Embed"]}.')
# get feature
if encoding == 'esm2_15b':
feature = ESM2_15b_EmbedFeaturizer(protein=protein_name, use_cache=True)
elif encoding == 'esm2_3b':
feature = ESM2EmbedFeaturizer(protein=protein_name, use_cache=True)
elif encoding == 'onehot':
feature = OneHotFeaturizer(protein=protein_name, use_cache=True)
elif encoding == 'georgiev':
feature = GeorgievFeaturizer(protein=protein_name, use_cache=True)
elif encoding == 'aaidx':
feature = AAIdxFeaturizer(protein=protein_name, use_cache=True)
elif encoding == 'onehot_and_esm2_15b':
feature = OnehotAndESM2_15bEmbedFeaturizer(protein=protein_name, use_cache=True)
elif encoding == 'ankh_base':
feature = AnkhBaseFeaturizer(protein=protein_name, use_cache=True, batch_size=batch_size)
elif encoding == 'ankh_large':
feature = AnkhLargeFeaturizer(protein=protein_name, use_cache=True, batch_size=batch_size)
elif encoding == 'ProtT5_XL_U50_Embed':
feature = ProtT5_XL_U50_EmbedFeaturizer(protein=protein_name, use_cache=True, batch_size=batch_size)
return feature |