output/preprocess/Cervical_Cancer/code/TCGA.py

# Path Configuration
from tools.preprocess import *

# Processing context
trait = "Cervical_Cancer"

# Input paths
tcga_root_dir = "../DATA/TCGA"

# Output paths
out_data_file = "./output/z2/preprocess/Cervical_Cancer/TCGA.csv"
out_gene_data_file = "./output/z2/preprocess/Cervical_Cancer/gene_data/TCGA.csv"
out_clinical_data_file = "./output/z2/preprocess/Cervical_Cancer/clinical_data/TCGA.csv"
json_path = "./output/z2/preprocess/Cervical_Cancer/cohort_info.json"


# Step 1: Initial Data Loading
import os
import re
import pandas as pd

# 1) Select the most relevant TCGA cohort directory for the trait
all_entries = os.listdir(tcga_root_dir)
subdirs = [d for d in all_entries if os.path.isdir(os.path.join(tcga_root_dir, d))]

# Prefer names containing "cervic" or "cesc"
pattern = re.compile(r'(cervic|cesc)', re.IGNORECASE)
matches = [d for d in subdirs if pattern.search(d)]

selected_cohort_dir = None
if matches:
    # Prefer the most specific match containing "cervical" first, else pick the first match
    cervical_matches = [d for d in matches if re.search(r'cervical', d, re.IGNORECASE)]
    selected = cervical_matches[0] if cervical_matches else matches[0]
    selected_cohort_dir = os.path.join(tcga_root_dir, selected)

if selected_cohort_dir is None:
    # No suitable directory found; record and skip
    validate_and_save_cohort_info(
        is_final=False,
        cohort="TCGA",
        info_path=json_path,
        is_gene_available=False,
        is_trait_available=False
    )
    tcga_skip_trait = True
else:
    tcga_skip_trait = False

# 2) Identify clinical and genetic file paths
clinical_file_path = None
genetic_file_path = None

if not tcga_skip_trait:
    try:
        clinical_file_path, genetic_file_path = tcga_get_relevant_filepaths(selected_cohort_dir)
    except Exception:
        # Fallback manual search if helper fails
        files = os.listdir(selected_cohort_dir)
        clinical_candidates = [f for f in files if 'clinicalmatrix' in f.lower()]
        genetic_candidates = [f for f in files if 'pancan' in f.lower()]
        if clinical_candidates and genetic_candidates:
            clinical_file_path = os.path.join(selected_cohort_dir, clinical_candidates[0])
            genetic_file_path = os.path.join(selected_cohort_dir, genetic_candidates[0])
        else:
            validate_and_save_cohort_info(
                is_final=False,
                cohort="TCGA",
                info_path=json_path,
                is_gene_available=bool(genetic_candidates),
                is_trait_available=bool(clinical_candidates)
            )
            tcga_skip_trait = True

# 3) Load both files as DataFrames
tcga_clinical_df = None
tcga_genetic_df = None

if not tcga_skip_trait:
    tcga_clinical_df = pd.read_csv(clinical_file_path, sep='\t', index_col=0, low_memory=False)
    tcga_genetic_df = pd.read_csv(genetic_file_path, sep='\t', index_col=0, low_memory=False)

    # 4) Print the column names of the clinical data
    print(list(tcga_clinical_df.columns))

# Step 2: Find Candidate Demographic Features
import os
import re
import pandas as pd

# The list of column names from the previous step
previous_columns = ['_INTEGRATION', '_PATIENT', '_cohort', '_primary_disease', '_primary_site', 'additional_pharmaceutical_therapy', 'additional_radiation_therapy', 'additional_treatment_completion_success_outcome', 'adjuvant_rad_therapy_prior_admin', 'age_at_initial_pathologic_diagnosis', 'age_began_smoking_in_years', 'agent_total_dose_count', 'assessment_timepoint_category', 'bcr_followup_barcode', 'bcr_patient_barcode', 'bcr_sample_barcode', 'birth_control_pill_history_usage_category', 'brachytherapy_administered_status', 'brachytherapy_first_reference_point_administered_total_dose', 'brachytherapy_method_other_specify_text', 'brachytherapy_method_type', 'cervical_carcinoma_corpus_uteri_involvement_indicator', 'cervical_carcinoma_pelvic_extension_text', 'cervical_neoplasm_pathologic_margin_involved_text', 'cervical_neoplasm_pathologic_margin_involved_type', 'chemotherapy_negation_radiation_therapy_concurrent_adminstrd_txt', 'chemotherapy_negation_radiation_therapy_concurrnt_nt_dmnstrd_rsn', 'chemotherapy_regimen_type', 'clinical_stage', 'concurrent_chemotherapy_dose', 'days_to_birth', 'days_to_brachytherapy_begin_occurrence', 'days_to_brachytherapy_end_occurrence', 'days_to_chemotherapy_end', 'days_to_chemotherapy_start', 'days_to_collection', 'days_to_death', 'days_to_diagnostic_computed_tomography_performed', 'days_to_diagnostic_mri_performed', 'days_to_fdg_or_ct_pet_performed', 'days_to_initial_pathologic_diagnosis', 'days_to_last_followup', 'days_to_new_tumor_event_additional_surgery_procedure', 'days_to_new_tumor_event_after_initial_treatment', 'days_to_performance_status_assessment', 'days_to_radiation_therapy_end', 'days_to_radiation_therapy_start', 'death_cause_text', 'diagnostic_ct_result_outcome', 'diagnostic_mri_result_outcome', 'dose_frequency_text', 'eastern_cancer_oncology_group', 'ectopic_pregnancy_count', 'external_beam_radiation_therapy_administered_status', 'external_beam_radiation_therapy_administrd_prrtc_rgn_lymph_nd_ds', 'fdg_or_ct_pet_performed_outcome', 'female_breast_feeding_or_pregnancy_status_indicator', 'followup_case_report_form_submission_reason', 'form_completion_date', 'gender', 'height', 'histological_type', 'history_of_neoadjuvant_treatment', 'human_papillomavirus_laboratory_procedure_performed_name', 'human_papillomavirus_laboratory_procedure_performed_text', 'human_papillomavirus_other_type_text', 'human_papillomavirus_type', 'hysterectomy_performed_text', 'hysterectomy_performed_type', 'icd_10', 'icd_o_3_histology', 'icd_o_3_site', 'informed_consent_verified', 'init_pathology_dx_method_other', 'initial_pathologic_diagnosis_method', 'initial_weight', 'is_ffpe', 'keratinizing_squamous_cell_carcinoma_present_indicator', 'lost_follow_up', 'lymph_node_examined_count', 'lymph_node_location_positive_pathology_name', 'lymph_node_location_positive_pathology_text', 'lymphovascular_invasion_indicator', 'menopause_status', 'neoplasm_histologic_grade', 'new_neoplasm_event_occurrence_anatomic_site', 'new_neoplasm_event_post_initial_therapy_diagnosis_method_text', 'new_neoplasm_event_post_initial_therapy_diagnosis_method_type', 'new_neoplasm_event_type', 'new_neoplasm_occurrence_anatomic_site_text', 'new_tumor_event_additional_surgery_procedure', 'new_tumor_event_after_initial_treatment', 'number_of_lymphnodes_positive_by_he', 'number_of_lymphnodes_positive_by_ihc', 'number_of_successful_pregnancies_which_resultd_n_t_lst_1_lv_brth', 'number_pack_years_smoked', 'oct_embedded', 'oligonucleotide_primer_pair_laboratory_procedure_performed_name', 'other_chemotherapy_agent_administration_specify', 'other_dx', 'pathologic_M', 'pathologic_N', 'pathologic_T', 'pathology_report_file_name', 'patient_death_reason', 'patient_history_immune_system_and_related_disorders_name', 'patient_history_immune_system_and_related_disorders_text', 'patient_id', 'patient_pregnancy_spontaneous_abortion_count', 'patient_pregnancy_therapeutic_abortion_count', 'performance_status_scale_timing', 'person_neoplasm_cancer_status', 'postoperative_rx_tx', 'pregnancy_stillbirth_count', 'primary_lymph_node_presentation_assessment', 'primary_therapy_outcome_success', 'radiation_therapy', 'radiation_therapy_not_administered_reason', 'radiation_therapy_not_administered_specify', 'radiation_type_notes', 'residual_disease_post_new_tumor_event_margin_status', 'rt_administered_type', 'rt_pelvis_administered_total_dose', 'sample_type', 'sample_type_id', 'standardized_uptake_value_cervix_uteri_assessment_measurement', 'stopped_smoking_year', 'system_version', 'targeted_molecular_therapy', 'tissue_prospective_collection_indicator', 'tissue_retrospective_collection_indicator', 'tissue_source_site', 'tobacco_smoking_history', 'total_number_of_pregnancies', 'tumor_response_cdus_type', 'tumor_tissue_site', 'vial_number', 'vital_status', 'weight', 'year_of_initial_pathologic_diagnosis', '_GENOMIC_ID_TCGA_CESC_exp_HiSeqV2_exon', '_GENOMIC_ID_TCGA_CESC_miRNA_HiSeq', '_GENOMIC_ID_TCGA_CESC_exp_HiSeqV2_PANCAN', '_GENOMIC_ID_data/public/TCGA/CESC/miRNA_HiSeq_gene', '_GENOMIC_ID_TCGA_CESC_PDMRNAseq', '_GENOMIC_ID_TCGA_CESC_RPPA', '_GENOMIC_ID_TCGA_CESC_hMethyl450', '_GENOMIC_ID_TCGA_CESC_mutation_bcgsc_gene', '_GENOMIC_ID_TCGA_CESC_exp_HiSeqV2_percentile', '_GENOMIC_ID_TCGA_CESC_mutation', '_GENOMIC_ID_TCGA_CESC_mutation_broad_gene', '_GENOMIC_ID_TCGA_CESC_mutation_ucsc_maf_gene', '_GENOMIC_ID_TCGA_CESC_mutation_curated_wustl_gene', '_GENOMIC_ID_TCGA_CESC_exp_HiSeqV2', '_GENOMIC_ID_TCGA_CESC_gistic2', '_GENOMIC_ID_TCGA_CESC_PDMRNAseqCNV', '_GENOMIC_ID_TCGA_CESC_gistic2thd']

# Refined patterns for age and gender
age_pattern = re.compile(r'(^|[_\W])age([_\W]|$)')
birth_pattern = re.compile(r'(^|[_\W])(days_to_birth|year_of_birth|date_of_birth|birth_year)([_\W]|$)')
gender_pattern = re.compile(r'(^|[_\W])gender([_\W]|$)|(^|[_\W])sex([_\W]|$)')

candidate_age_cols = []
candidate_gender_cols = []

for col in previous_columns:
    low = col.lower()
    if (age_pattern.search(low) or birth_pattern.search(low)) and ('birth_control' not in low and 'stillbirth' not in low):
        candidate_age_cols.append(col)
    if gender_pattern.search(low):
        candidate_gender_cols.append(col)

# Print the required lists in the specified format
print(f"candidate_age_cols = {candidate_age_cols}")
print(f"candidate_gender_cols = {candidate_gender_cols}")

# Load clinical data and preview candidate columns if available
clinical_df = None
try:
    cohort_dir = None
    for entry in os.scandir(tcga_root_dir):
        if entry.is_dir() and 'CESC' in entry.name.upper():
            cohort_dir = entry.path
            break
    if cohort_dir is None:
        for root, dirs, _ in os.walk(tcga_root_dir):
            for d in dirs:
                if 'CESC' in d.upper():
                    cohort_dir = os.path.join(root, d)
                    break
            if cohort_dir is not None:
                break

    if cohort_dir is not None:
        clinical_file_path, _ = tcga_get_relevant_filepaths(cohort_dir)
        clinical_df = pd.read_csv(clinical_file_path, sep='\t', index_col=0, dtype=str)
except Exception:
    clinical_df = None

# Extract and preview
age_preview = {}
gender_preview = {}

if isinstance(clinical_df, pd.DataFrame):
    age_cols_existing = [c for c in candidate_age_cols if c in clinical_df.columns]
    gender_cols_existing = [c for c in candidate_gender_cols if c in clinical_df.columns]

    if age_cols_existing:
        age_preview = preview_df(clinical_df[age_cols_existing])
    if gender_cols_existing:
        gender_preview = preview_df(clinical_df[gender_cols_existing])

print("age_preview =", age_preview)
print("gender_preview =", gender_preview)

# Step 3: Select Demographic Features
# Select columns based on preview and typical TCGA conventions
age_col = 'age_at_initial_pathologic_diagnosis' if 'age_at_initial_pathologic_diagnosis' in candidate_age_cols else None
gender_col = 'gender' if 'gender' in candidate_gender_cols else None

# Explicitly print chosen columns and their previews
print("Selected age_col:", age_col)
print("Preview of selected age_col:", age_preview.get(age_col) if age_col else None)

print("Selected gender_col:", gender_col)
print("Preview of selected gender_col:", gender_preview.get(gender_col) if gender_col else None)

# Step 4: Feature Engineering and Validation
import os
import pandas as pd

# 1) Extract and standardize clinical features (trait, Age, Gender)
selected_clinical_df = tcga_select_clinical_features(
    clinical_df=tcga_clinical_df,
    trait=trait,
    age_col=age_col,
    gender_col=gender_col
)

# 2) Normalize gene symbols and save normalized gene data
def _looks_like_sample_ids(idx):
    try:
        return any(str(i).startswith('TCGA-') for i in list(idx)[:10])
    except Exception:
        return any(str(i).startswith('TCGA-') for i in idx)

gene_df = tcga_genetic_df
# Ensure genes are in index for normalization
if _looks_like_sample_ids(gene_df.index):
    gene_df = gene_df.T

normalized_gene_df = normalize_gene_symbols_in_index(gene_df)

# Optional post-check to ensure samples are columns
if not _looks_like_sample_ids(normalized_gene_df.columns) and _looks_like_sample_ids(normalized_gene_df.index):
    normalized_gene_df = normalized_gene_df.T

# Save normalized gene data
os.makedirs(os.path.dirname(out_gene_data_file), exist_ok=True)
normalized_gene_df.to_csv(out_gene_data_file)

# 3) Link clinical and genetic data on intersecting sample IDs
common_samples = selected_clinical_df.index.intersection(normalized_gene_df.columns)
linked_clinical = selected_clinical_df.loc[common_samples]
linked_gene = normalized_gene_df[common_samples].T  # samples x genes
linked_data = pd.concat([linked_clinical, linked_gene], axis=1)

# 4) Handle missing values
linked_data = handle_missing_values(linked_data, trait_col=trait)

# 5) Determine severe bias; remove biased demographics
trait_biased, linked_data = judge_and_remove_biased_features(linked_data, trait=trait)

# 6) Final validation and save cohort info
covariate_cols = [trait, 'Age', 'Gender']
gene_cols_after = [c for c in linked_data.columns if c not in covariate_cols]
is_gene_available = bool(len(gene_cols_after) > 0)
is_trait_available = bool((trait in linked_data.columns) and bool(linked_data[trait].notna().any()))

note_parts = [
    f"INFO: Cohort TCGA CESC. Samples linked: {len(linked_data)}.",
    f"INFO: Gene features retained: {len(gene_cols_after)}.",
    f"INFO: Age included: {'Age' in linked_data.columns}.",
    f"INFO: Gender included: {'Gender' in linked_data.columns}.",
]
note = " ".join(note_parts)

is_usable = validate_and_save_cohort_info(
    is_final=True,
    cohort="TCGA",
    info_path=json_path,
    is_gene_available=is_gene_available,
    is_trait_available=is_trait_available,
    is_biased=bool(trait_biased),
    df=linked_data,
    note=note
)

# 7) Save linked data only if usable
if is_usable:
    os.makedirs(os.path.dirname(out_data_file), exist_ok=True)
    linked_data.to_csv(out_data_file)