NDA and dbGaP data submission SOP

This repository documents and develops a reproducible workflow for preparing LIBD controlled-access data submissions to the NIMH Data Archive (NDA), with dbGaP registration/linkage tracked where required for human genomic data.

The first worked example is SNP-array genotype data preparation for the union of:

• SCZ-PNN donors represented by NDA Study 3054
• MBv donors represented by Ryan's 119-row MBv donor/sample list
• overlapping donors/controls deduplicated before packaging

Current goal

Prepare, validate, and submit a genotype data package to NDA collection C5229, while documenting the process well enough to become a reusable SOP.

Parallel discovery work:

• understand Ryan Miller's existing NDA data packaging work for collection C5229
• inspect whether MBv Visium prep exists on JHPCE for experiment 2863
• prepare genotype inputs on srv16 from the documented genotype locations
• determine how the NDA genotype submission should link to NDA Studies, Experiments, Submissions, and the dbGaP record

Documentation map

• docs/NDA-collection-info.md - current interpretation of collection C5229, including Studies, Experiments, Submissions, counts, and open mapping questions.
• docs/workstreams.md - live tracker for active workstreams, blockers, evidence, and next actions.
• docs/sop-nda-genotype-submission.md - discovery-stage SOP for NDA genotype package preparation, validation, and submission.
• docs/sop-nda-genotype-data-package.md - neutral command-level guide for preparing an NDA genotype package from VCF inputs and donor/GUID metadata.
• docs/handoffs/jhpce-srv16-nda-genotype-handoff.md - handoff for JHPCE Ryan/MBv package discovery and srv16 genotype package preparation.

Controlled data rules

Do not commit controlled donor-level or genotype data.

Do not commit:

• donor rows
• BrNum row lists, unless explicitly approved
• GUID or pseudo-GUID crosswalks
• NDA downloads
• genotype files
• local validation outputs containing donor identifiers
• archive files or package payloads

Safe to commit:

• SOP documentation
• aggregate counts
• non-sensitive command patterns
• high-level path references needed for reproducibility
• notes that explicitly avoid donor rows and GUID mappings

BrNums may be used in local work or chat when needed.

Generated local working directories for controlled files should stay untracked.