#!/bin/bash -Ceuo pipefail
SEX=$(python3 -c "
import pandas as pd
sex = pd.read_csv('sex_calls.csv')
lookup = dict(zip(sex['specimen_id'], sex['inferred_sex']))
s = lookup.get('custom_Sig_18_tumor_normal', 'unknown')
print('male' if s != 'female' else 'female')
")

python /usr/local/bin/normalize_sample.py \
    --counts Sig_18_tissue_probes.tsv.gz \
    --batch-reference batch_reference.parquet \
    --gc-reference /usr/local/share/probe_gc_content.tsv \
    --sex $SEX \
    --sample-id Sig_18_tissue \
    --specimen-id custom_Sig_18_tumor_normal \
    --flowcell unknown \
    --sample-type tumor \
    --output Sig_18_tissue_normalized.parquet \
    --feature-stats-output Sig_18_tissue_gam_stats.json

cat <<-END_VERSIONS > versions.yml
"NFCORE_SAREK:SAREK:BAM_VARIANT_CALLING_SOMATIC_ALL:SOMATIC_CNV_CALLING:SOMA_CNV_NORMALIZE":
    soma-cnv: 20260604-914d02b
END_VERSIONS