File size: 14,020 Bytes
6b8ee1b | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 | # Path Configuration
from tools.preprocess import *
# Processing context
trait = "Depression"
cohort = "GSE201332"
# Input paths
in_trait_dir = "../DATA/GEO/Depression"
in_cohort_dir = "../DATA/GEO/Depression/GSE201332"
# Output paths
out_data_file = "./output/z2/preprocess/Depression/GSE201332.csv"
out_gene_data_file = "./output/z2/preprocess/Depression/gene_data/GSE201332.csv"
out_clinical_data_file = "./output/z2/preprocess/Depression/clinical_data/GSE201332.csv"
json_path = "./output/z2/preprocess/Depression/cohort_info.json"
# Step 1: Initial Data Loading
from tools.preprocess import *
# 1. Identify the paths to the SOFT file and the matrix file
soft_file, matrix_file = geo_get_relevant_filepaths(in_cohort_dir)
# 2. Read the matrix file to obtain background information and sample characteristics data
background_prefixes = ['!Series_title', '!Series_summary', '!Series_overall_design']
clinical_prefixes = ['!Sample_geo_accession', '!Sample_characteristics_ch1']
background_info, clinical_data = get_background_and_clinical_data(matrix_file, background_prefixes, clinical_prefixes)
# 3. Obtain the sample characteristics dictionary from the clinical dataframe
sample_characteristics_dict = get_unique_values_by_row(clinical_data)
# 4. Explicitly print out all the background information and the sample characteristics dictionary
print("Background Information:")
print(background_info)
print("Sample Characteristics Dictionary:")
print(sample_characteristics_dict)
# Step 2: Dataset Analysis and Clinical Feature Extraction
import re
import os
# 1) Gene Expression Data Availability
is_gene_available = True # "Transcriptional profiling" of whole blood for DEGs indicates mRNA expression data.
# 2) Variable Availability and Converters
trait_row = 1 # 'subject status: heathy controls' vs 'subject status: MDD patients'
age_row = 3 # 'age: 43y', etc.
gender_row = 2 # 'gender: male' / 'gender: female'
def _after_colon(val):
if val is None:
return None
s = str(val).strip()
if ':' in s:
s = s.split(':', 1)[1].strip()
return s
def convert_trait(val):
s = _after_colon(val)
if s is None or s == '':
return None
s_low = s.lower()
# Map MDD/depression to 1, controls/healthy to 0
if any(k in s_low for k in ['mdd', 'depress']):
return 1
if any(k in s_low for k in ['control', 'healthy', 'normal', 'hc']):
return 0
return None
def convert_age(val):
s = _after_colon(val)
if s is None or s == '':
return None
m = re.search(r'(\d+(\.\d+)?)', s)
if m:
num = float(m.group(1))
return num
return None
def convert_gender(val):
s = _after_colon(val)
if s is None or s == '':
return None
s_low = s.lower()
if s_low in ['male', 'm']:
return 1
if s_low in ['female', 'f']:
return 0
return None
# 3) Save Metadata (initial filtering)
is_trait_available = trait_row is not None
_ = validate_and_save_cohort_info(
is_final=False,
cohort=cohort,
info_path=json_path,
is_gene_available=is_gene_available,
is_trait_available=is_trait_available
)
# 4) Clinical Feature Extraction (only if clinical data is available)
if trait_row is not None:
selected_clinical_df = geo_select_clinical_features(
clinical_df=clinical_data,
trait=trait,
trait_row=trait_row,
convert_trait=convert_trait,
age_row=age_row,
convert_age=convert_age,
gender_row=gender_row,
convert_gender=convert_gender
)
preview = preview_df(selected_clinical_df)
print(preview)
os.makedirs(os.path.dirname(out_clinical_data_file), exist_ok=True)
selected_clinical_df.to_csv(out_clinical_data_file, index=True)
# Step 3: Gene Data Extraction
# 1. Use the get_genetic_data function from the library to get the gene_data from the matrix_file previously defined.
gene_data = get_genetic_data(matrix_file)
# 2. Print the first 20 row IDs (gene or probe identifiers) for future observation.
print(gene_data.index[:20])
# Step 4: Gene Identifier Review
# The observed identifiers are numeric (e.g., '1', '2', ...), consistent with Entrez Gene IDs, not human gene symbols.
requires_gene_mapping = True
print(f"requires_gene_mapping = {requires_gene_mapping}")
# Step 5: Gene Annotation
# 1. Use the 'get_gene_annotation' function from the library to get gene annotation data from the SOFT file.
gene_annotation = get_gene_annotation(soft_file)
# 2. Use the 'preview_df' function from the library to preview the data and print out the results.
print("Gene annotation preview:")
print(preview_df(gene_annotation))
# Step 6: Gene Identifier Mapping
# Determine the probe ID column and candidate gene symbol columns
probe_col = 'ID' if 'ID' in gene_annotation.columns else None
if probe_col is None:
raise ValueError("Probe ID column 'ID' was not found in the gene annotation dataframe.")
# Candidate columns that may contain gene symbols or descriptions from which symbols can be extracted
candidate_gene_cols = [
'GENE_SYMBOL', 'Gene Symbol', 'Symbol', 'SYMBOL', 'Gene', 'GENE',
'GENE_NAME', 'Gene Name', 'GENE_TITLE', 'GENE TITLE', 'GENE_SYMBOLS',
'DESCRIPTION', 'DEFINITION', 'Product', 'PRODUCT', 'RefSeq', 'REFSEQ',
'ENTREZ_GENE_ID', 'ENTREZID', 'GB_ACC', 'SEQ_ACC', 'ORF', 'ACCNUM',
'SPOT_ID', 'NAME', 'SEQUENCE', 'CHROMOSOMAL_LOCATION'
]
present_gene_cols = [c for c in candidate_gene_cols if c in gene_annotation.columns]
if not present_gene_cols:
# Fallback: try any non-ID textual columns
present_gene_cols = [c for c in gene_annotation.columns if c != probe_col]
# Score candidate columns by how many rows yield at least one human gene symbol
best_col = None
best_count = -1
for c in present_gene_cols:
try:
tmp_map = get_gene_mapping(gene_annotation, prob_col=probe_col, gene_col=c)
except Exception:
continue
if tmp_map.empty:
continue
# Restrict to probes present in the expression data
tmp_map = tmp_map[tmp_map['ID'].isin(gene_data.index)]
if tmp_map.empty:
continue
# Count rows with at least one extracted human gene symbol
count_nonempty = tmp_map['Gene'].apply(extract_human_gene_symbols).apply(lambda x: len(x) if isinstance(x, list) else 0).gt(0).sum()
if count_nonempty > best_count:
best_count = count_nonempty
best_col = c
if best_col is None or best_count <= 0:
# As a last resort, use 'NAME' if available, otherwise raise an error
if 'NAME' in gene_annotation.columns:
best_col = 'NAME'
else:
raise ValueError("Could not identify a suitable annotation column containing gene symbols.")
# Build final mapping using the selected column
mapping_df = get_gene_mapping(gene_annotation, prob_col=probe_col, gene_col=best_col)
# Convert probe-level data to gene-level expression using the mapping
gene_data = apply_gene_mapping(expression_df=gene_data, mapping_df=mapping_df)
# Step 7: Data Normalization and Linking
import os
# 1. Normalize gene symbols and save
normalized_gene_data = normalize_gene_symbols_in_index(gene_data)
os.makedirs(os.path.dirname(out_gene_data_file), exist_ok=True)
normalized_gene_data.to_csv(out_gene_data_file)
# 2. Link clinical and genetic data (fix variable name)
linked_data = geo_link_clinical_genetic_data(selected_clinical_df, normalized_gene_data)
# 3. Handle missing values
linked_data = handle_missing_values(linked_data, trait)
# 4. Bias assessment and removal of biased demographics
is_trait_biased, unbiased_linked_data = judge_and_remove_biased_features(linked_data, trait)
# 5. Final validation and metadata
is_usable = validate_and_save_cohort_info(
True,
cohort,
json_path,
True,
True,
is_trait_biased,
unbiased_linked_data,
note="INFO: Probes mapped to symbols via annotation; symbols normalized using NCBI synonyms."
)
# 6. Save linked data if usable
if is_usable:
os.makedirs(os.path.dirname(out_data_file), exist_ok=True)
unbiased_linked_data.to_csv(out_data_file)
# Step 8: Gene Identifier Mapping
import json
import re
# Reload raw expression data to ensure probe IDs
raw_expression_df = get_genetic_data(matrix_file)
# 1) Identify identifier column in annotation
probe_col = 'ID' if 'ID' in gene_annotation.columns else None
if probe_col is None:
raise ValueError("Probe ID column 'ID' was not found in the gene annotation dataframe.")
# Probes present in expression data
expr_probe_ids = set(raw_expression_df.index.astype(str))
# Exclude known control probes if CONTROL_TYPE is present
if 'CONTROL_TYPE' in gene_annotation.columns:
control_flags = gene_annotation['CONTROL_TYPE'].astype(str).str.lower()
non_control_mask = ~control_flags.isin(['pos', 'neg', 'control', 'empty', 'ignore'])
non_control_ids = set(gene_annotation.loc[non_control_mask, probe_col].astype(str))
else:
non_control_ids = set(gene_annotation[probe_col].astype(str))
valid_probe_ids = expr_probe_ids.intersection(non_control_ids)
# 2) Select the best annotation column containing gene symbols/descriptors
candidate_gene_cols = [
'GENE_SYMBOL', 'Gene Symbol', 'Symbol', 'SYMBOL', 'Gene', 'GENE',
'GENE_NAME', 'Gene Name', 'GENE_TITLE', 'GENE TITLE', 'GENE_SYMBOLS',
'DESCRIPTION', 'DEFINITION', 'Product', 'PRODUCT', 'RefSeq', 'REFSEQ',
'ENTREZ_GENE_ID', 'ENTREZID', 'GB_ACC', 'SEQ_ACC', 'ORF', 'ACCNUM',
'NAME', 'SEQUENCE', 'SPOT_ID', 'CHROMOSOMAL_LOCATION'
]
present_gene_cols = [c for c in candidate_gene_cols if c in gene_annotation.columns]
if not present_gene_cols:
present_gene_cols = [c for c in gene_annotation.columns if c != probe_col]
# Load synonym dictionary to score columns
with open("./metadata/gene_synonym.json", "r") as f:
synonym_dict = json.load(f)
synonym_keys = set(synonym_dict.keys())
# Token exclusion patterns (spike-ins, controls, generic RNA placeholders)
exclude_exact = {"GE_BRIGHTCORNER", "DARKCORNER", "EMPTY", "CONTROL", "NEG", "POS"}
exclude_regex = [
re.compile(r'^ERCC[\w-]*$', re.IGNORECASE),
re.compile(r'^RNA\d+$', re.IGNORECASE),
re.compile(r'^RNA\d+-\d+$', re.IGNORECASE),
re.compile(r'^NEG[\w-]*$', re.IGNORECASE),
re.compile(r'^POS[\w-]*$', re.IGNORECASE),
]
def filter_tokens(tokens):
kept = []
for t in tokens:
if not isinstance(t, str):
continue
u = t.upper()
if u in exclude_exact:
continue
if any(rx.match(u) for rx in exclude_regex):
continue
# Only keep tokens recognized by synonym dictionary
if u in synonym_keys:
kept.append(u)
return kept
def score_column(col_name):
tmp_map = get_gene_mapping(gene_annotation, prob_col=probe_col, gene_col=col_name)
if tmp_map.empty:
return 0, set()
tmp_map = tmp_map[tmp_map['ID'].astype(str).isin(valid_probe_ids)]
if tmp_map.empty:
return 0, set()
extracted = tmp_map['Gene'].apply(extract_human_gene_symbols)
# Filter symbols
filtered_lists = extracted.apply(filter_tokens)
# Count unique recognized symbols
uniq_syms = set(sym for lst in filtered_lists if isinstance(lst, list) for sym in lst)
return len(uniq_syms), uniq_syms
# First pass: score all present columns
scores = {}
uniq_syms_by_col = {}
for c in present_gene_cols:
cnt, uniq = score_column(c)
scores[c] = cnt
uniq_syms_by_col[c] = uniq
# Choose the best column by recognized count
best_col = max(scores, key=lambda k: scores[k]) if scores else None
best_count = scores.get(best_col, 0) if best_col is not None else 0
# Enforce fallback strategy if no recognized symbols
if best_count <= 0:
for fallback in ['NAME', 'SEQUENCE']:
if fallback in gene_annotation.columns:
cnt, uniq = score_column(fallback)
if cnt > 0:
best_col = fallback
best_count = cnt
uniq_syms_by_col[best_col] = uniq
break
# As a safety, avoid SPOT_ID unless it yields recognized symbols
if (best_col is None) or (best_count <= 0) or (best_col == 'SPOT_ID' and best_count <= 0):
raise ValueError("Could not identify an annotation column that yields recognized human gene symbols.")
print(f"Selected identifier column: {probe_col}")
print(f"Selected gene annotation column: {best_col} (recognized_symbols={best_count})")
# 3) Build mapping and apply to convert probes -> genes, with explicit filtering
mapping_df_raw = get_gene_mapping(gene_annotation, prob_col=probe_col, gene_col=best_col)
mapping_df_raw = mapping_df_raw[mapping_df_raw['ID'].astype(str).isin(valid_probe_ids)].copy()
# Extract and filter tokens per row
extracted = mapping_df_raw['Gene'].apply(extract_human_gene_symbols)
filtered_tokens = extracted.apply(filter_tokens)
# Keep rows that have at least one recognized, non-excluded symbol
keep_mask = filtered_tokens.apply(lambda lst: isinstance(lst, list) and len(lst) > 0)
mapping_df_filtered = mapping_df_raw.loc[keep_mask, ['ID']].copy()
# Join tokens back to a single string so that apply_gene_mapping can re-extract correctly
mapping_df_filtered['Gene'] = filtered_tokens.loc[keep_mask].apply(lambda lst: ';'.join(lst))
print(f"Mapping dataframe shape after filtering: {mapping_df_filtered.shape}")
# Show a small sample of recognized symbols we will map
recognized_syms_sample = sorted(list(set(sym for lst in filtered_tokens.loc[keep_mask] for sym in lst)))[:15]
print(f"Sample of recognized symbols to be mapped: {recognized_syms_sample}")
if mapping_df_filtered.empty:
raise ValueError("Derived mapping_df is empty after filtering; cannot map probes to gene symbols.")
gene_data = apply_gene_mapping(expression_df=raw_expression_df, mapping_df=mapping_df_filtered)
print(f"Gene-level expression shape: {gene_data.shape}")
print(f"First 10 genes mapped: {list(gene_data.index[:10])}")
if gene_data.empty:
raise ValueError("Resulting gene_data is empty after applying mapping.") |