Downloading: s3://natera-rnd-pltf-dev-nextflow-scratch-01/work/50/65e46a4fbef72d4af4a01b6474c454/Sig_18_Blood.recal_probes.tsv.gz
  Downloading: s3://natera-rnd-pltf-dev-nextflow-scratch-01/work/13/8fc1994ddd635bd16ed81358dbd9b2/.command.sh
  Downloading: s3://natera-rnd-pltf-dev-nextflow-scratch-01/work/13/8fc1994ddd635bd16ed81358dbd9b2/.command.run
  Downloading: s3://natera-rnd-pltf-dev-nextflow-scratch-01/work/9d/32768b96495c76926383d03309e242/HCC1395_BL.recal_probes.tsv.gz
  Downloading: s3://natera-rnd-pltf-dev-nextflow-scratch-01/work/tmp/f5/7627a0596852561130d9e7cba0f9f0/manifest.tsv
==> STAGING COMPLETE (5 inputs)

Blood manifest: 2 samples
Loaded 9,638 rows from 2 blood samples in 0.0s
GC reference: 4,819 probes
  Pass 1: 4-feat blood GAM (16 workers)...
    2/2 blood samples, 4s, R²=0.8025
    Excluded 2 samples with chrX < 0.1 (data quality)
    Probe coverage: 4,819/4,819 valid probes
  Preliminary 5-feat GAM on blood (16 workers)...
    2/2 blood prelim 5-feat, 6s, R²=0.9587
    Probe mask: 237 masked (4.9%) — 80 low cov, 186 high var, 20 extreme
Wrote batch_reference.parquet (4,819 probes) to ./batch_reference.parquet
Wrote sex_calls.csv (2 samples) to ./sex_calls.csv
Wrote feature_stats.csv (2 samples) to ./feature_stats.csv
Build reference complete in 10.9s total