Liu-Hy's picture
Add files using upload-large-folder tool
75faa94 verified
# Path Configuration
from tools.preprocess import *
# Processing context
trait = "Eczema"
cohort = "GSE123086"
# Input paths
in_trait_dir = "../DATA/GEO/Eczema"
in_cohort_dir = "../DATA/GEO/Eczema/GSE123086"
# Output paths
out_data_file = "./output/preprocess/3/Eczema/GSE123086.csv"
out_gene_data_file = "./output/preprocess/3/Eczema/gene_data/GSE123086.csv"
out_clinical_data_file = "./output/preprocess/3/Eczema/clinical_data/GSE123086.csv"
json_path = "./output/preprocess/3/Eczema/cohort_info.json"
# Get paths to the SOFT and matrix files
soft_file, matrix_file = geo_get_relevant_filepaths(in_cohort_dir)
# Get background info and clinical data from matrix file
background_info, clinical_data = get_background_and_clinical_data(matrix_file)
# Get unique values for each feature (row) in clinical data
unique_values_dict = get_unique_values_by_row(clinical_data)
# Print background info
print("=== Dataset Background Information ===")
print(background_info)
print("\n=== Sample Characteristics ===")
print(json.dumps(unique_values_dict, indent=2))
# 1. Gene Expression Data Availability
# Based on background info, this is a microarray gene expression study
is_gene_available = True
# 2. Data Type Conversion Functions
def convert_trait(x):
if pd.isna(x):
return None
# Extract value after colon and convert to binary
value = x.split(": ")[1].strip()
if value == "ATOPIC_ECZEMA":
return 1
elif value == "HEALTHY_CONTROL":
return 0
return None
def convert_age(x):
if pd.isna(x):
return None
try:
# Extract numeric age value after colon
age = int(x.split(": ")[1])
return age
except:
return None
def convert_gender(x):
if pd.isna(x):
return None
# Extract value after colon and convert to binary
value = x.split(": ")[1].strip()
if value.upper() == "FEMALE":
return 0
elif value.upper() == "MALE":
return 1
return None
# Find data rows in sample characteristics
trait_row = 1 # Primary diagnosis in row 1
age_row = 3 # Age appears in rows 3 and 4, but row 3 has more entries
gender_row = 2 # Sex information in row 2
# 3. Save metadata for initial filtering
validate_and_save_cohort_info(
is_final=False,
cohort=cohort,
info_path=json_path,
is_gene_available=is_gene_available,
is_trait_available=(trait_row is not None)
)
# 4. Extract clinical features
clinical_df = pd.DataFrame(clinical_data)
selected_clinical_df = geo_select_clinical_features(
clinical_df,
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 and save clinical data
print("Preview of extracted clinical features:")
print(preview_df(selected_clinical_df))
selected_clinical_df.to_csv(out_clinical_data_file)
# Extract gene expression data from matrix file
genetic_df = get_genetic_data(matrix_file)
# Print DataFrame shape and first 20 row IDs
print("DataFrame shape:", genetic_df.shape)
print("\nFirst 20 row IDs:")
print(genetic_df.index[:20])
print("\nPreview of first few rows and columns:")
print(genetic_df.head().iloc[:, :5])
# Examine gene identifiers - these appear to be numbers rather than standard gene symbols
# Numbers indicate probe or probe set IDs from a microarray platform
# Will need to be mapped to gene symbols
requires_gene_mapping = True
# Extract probe IDs and gene name mapping from SOFT file
gene_metadata = get_gene_annotation(soft_file)
# Print field information to check available gene identifiers
print("Sample of probe ID field:")
print(gene_metadata['ID'].head())
print("\nAll column names:")
print(list(gene_metadata.columns))
print("\nSample of gene metadata rows:")
pd.set_option('display.max_columns', None)
print(gene_metadata.head())
# Looking at the gene metadata, let's try to extract more information from the SOFT file
# Extract probe IDs and gene name mapping from SOFT file again, but this time don't filter out comment lines
gene_metadata = pd.read_csv(soft_file, compression='gzip', sep='\t', comment=None, on_bad_lines='skip')
# Get relevant columns for mapping
id_col = [col for col in gene_metadata.columns if 'ID_REF' in col or 'ID' in col][0]
gene_col = [col for col in gene_metadata.columns if 'GENE_SYMBOL' in col][0]
# Create mapping dataframe
mapping_df = gene_metadata[[id_col, gene_col]].copy()
mapping_df.columns = ['ID', 'Gene']
mapping_df = mapping_df.dropna()
# Apply the mapping to convert probe-level data to gene-level data
gene_data = apply_gene_mapping(genetic_df, mapping_df)
print("Gene data shape after mapping:", gene_data.shape)
print("\nPreview of first few genes and samples:")
print(gene_data.head().iloc[:, :5])
# Save the gene expression data
gene_data.to_csv(out_gene_data_file)
# Extract gene annotation data using the library function
gene_metadata = get_gene_annotation(soft_file)
# Get column with ENTREZ_GENE_IDs to map to NCBI gene symbols
mapping_df = get_gene_mapping(gene_metadata, 'ID', 'ENTREZ_GENE_ID')
# Apply the mapping to convert probe-level data to gene-level data
gene_data = apply_gene_mapping(genetic_df, mapping_df)
print("Gene data shape after mapping:", gene_data.shape)
print("\nPreview of first few genes and samples:")
print(gene_data.head().iloc[:, :5])
# Save the gene expression data
gene_data.to_csv(out_gene_data_file)
# Skip gene symbol normalization since we have no valid gene data
gene_data = pd.read_csv(out_gene_data_file, index_col=0)
if len(gene_data) == 0:
# Load clinical data for validation
clinical_df = pd.read_csv(out_clinical_data_file, index_col=0)
# Check for biased features in clinical data
trait_biased, clinical_df = judge_and_remove_biased_features(clinical_df, trait)
validate_and_save_cohort_info(
is_final=True,
cohort=cohort,
info_path=json_path,
is_gene_available=True, # Genes exist but mapping failed
is_trait_available=True,
is_biased=trait_biased,
df=clinical_df,
note="Gene mapping failed - no valid gene expression data produced"
)
else:
# Original processing steps
gene_data = normalize_gene_symbols_in_index(gene_data)
os.makedirs(os.path.dirname(out_gene_data_file), exist_ok=True)
gene_data.to_csv(out_gene_data_file)
# Link clinical and genetic data
clinical_df = pd.read_csv(out_clinical_data_file, index_col=0)
linked_data = geo_link_clinical_genetic_data(clinical_df, gene_data)
# Handle missing values
linked_data = handle_missing_values(linked_data, trait)
# Check for biased features
trait_biased, linked_data = judge_and_remove_biased_features(linked_data, trait)
# Final validation
is_usable = validate_and_save_cohort_info(
is_final=True,
cohort=cohort,
info_path=json_path,
is_gene_available=True,
is_trait_available=True,
is_biased=trait_biased,
df=linked_data,
note="Study comparing Eczema patient vs healthy control gene expression in CD4+ T cells"
)
# Save linked data if usable
if is_usable:
os.makedirs(os.path.dirname(out_data_file), exist_ok=True)
linked_data.to_csv(out_data_file)
# Get paths to the SOFT and matrix files
soft_file, matrix_file = geo_get_relevant_filepaths(in_cohort_dir)
# Get background info and clinical data from matrix file
background_info, clinical_data = get_background_and_clinical_data(matrix_file)
# Get unique values for each feature (row) in clinical data
unique_values_dict = get_unique_values_by_row(clinical_data)
# Print background info
print("=== Dataset Background Information ===")
print(background_info)
print("\n=== Sample Characteristics ===")
print(json.dumps(unique_values_dict, indent=2))
# 1. Gene Expression Data Availability
# Based on the series description mentioning gene expression microarray analysis, RNA extraction,
# and Agilent microarray processing, this dataset contains gene expression data
is_gene_available = True
# 2.1 Data Availability
# Trait (primary diagnosis) is in row 1
trait_row = 1
# Gender is in row 3 (and partly in row 2)
gender_row = 3
# Age appears in rows 3 and 4
age_row = 3
# 2.2 Data Type Conversion Functions
def convert_trait(value: str) -> int:
"""Convert trait values to binary (0: control, 1: case)"""
if not isinstance(value, str):
return None
value = value.split(': ')[-1].strip().upper()
if "ATOPIC_ECZEMA" in value:
return 1
elif "HEALTHY_CONTROL" in value:
return 0
return None
def convert_age(value: str) -> float:
"""Convert age values to continuous numbers"""
if not isinstance(value, str) or not value.startswith('age: '):
return None
try:
return float(value.split(': ')[1])
except:
return None
def convert_gender(value: str) -> int:
"""Convert gender values to binary (0: female, 1: male)"""
if not isinstance(value, str) or not value.startswith('Sex: '):
return None
value = value.split(': ')[1].strip().upper()
if value == 'FEMALE':
return 0
elif value == 'MALE':
return 1
return None
# 3. Save Metadata
validate_and_save_cohort_info(
is_final=False,
cohort=cohort,
info_path=json_path,
is_gene_available=is_gene_available,
is_trait_available=(trait_row is not None)
)
# 4. Extract Clinical Features
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 and save clinical data
print("Preview of extracted clinical features:")
print(preview_df(selected_clinical_df))
selected_clinical_df.to_csv(out_clinical_data_file)
# Based on the preview data, we can see this dataset has trait data (0s and 1s), age data (numeric age values), and gender data (0s and 1s)
is_gene_available = True # This is a GEO dataset so likely contains gene expression data
# Define conversion functions
def convert_trait(value):
if pd.isna(value):
return None
try:
val = float(value.split(":")[-1].strip() if ":" in value else value)
return val # Already binary (0/1) format
except:
return None
def convert_age(value):
if pd.isna(value):
return None
try:
val = float(value.split(":")[-1].strip() if ":" in value else value)
return val
except:
return None
def convert_gender(value):
if pd.isna(value):
return None
try:
val = float(value.split(":")[-1].strip() if ":" in value else value)
return val # Already in binary format where 1=male, 0=female
except:
return None
# Identify row indices for each variable based on the data preview
trait_row = 0
age_row = 1
gender_row = 2
# Initial filtering and save metadata
validate_and_save_cohort_info(is_final=False,
cohort=cohort,
info_path=json_path,
is_gene_available=is_gene_available,
is_trait_available=trait_row is not None)
# Extract clinical features since trait 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 the extracted features
print("\nPreview of extracted clinical features:")
print(preview_df(selected_clinical_df))
# Save clinical data
selected_clinical_df.to_csv(out_clinical_data_file)