[Week 8–10] AUD clinical validation: HCP S1200 discordant twin attractor analysis

[TheRoy666]

AUDPipeline is implemented and validated on synthetic cohorts. This issue
tracks deployment on real HCP S1200 discordant monozygotic twin data.

Tasks:

• Confirm HCP S1200 DUA covers rsfMRI data for AUD analysis
• Identify discordant MZ twin pairs in HCP S1200 (AUD-affected vs co-twin)
• Define reward network parcels in Schaefer-400 atlas (NAcc, caudate, OFC, vmPFC)
• Run AUDPipeline.run_cohort() on full twin cohort
• Run twin_discordance_analysis() — compute within-pair delta_E
• Generate Figure: group energy violin + twin pair delta_E bar chart

Primary metric: delta_E = E*(AUD) - E*(Control) for craving_to_rest, T=10 TRs.
This is the primary midterm evaluation deliverable (~July 11).

Tracks GSoC Weeks 8–10 deliverable.