REST-meta-MDD Skill (Dataset-Orchestration Layer)

Overview

rest-mneta-mdd-skill is the NeuroClaw orchestration skill for the REST-meta-MDD (Resting-State Meta-Major Depressive Disorder) dataset, a large-scale multi-site consortium project pooling resting-state fMRI data from 17 research sites across China.

It strictly follows the NeuroClaw hierarchical design principles:

• This skill only describes WHAT needs to be done and which tool skill to delegate to.
• It contains no implementation code or concrete commands.
• All concrete execution is delegated to existing base/tool skills via claw-shell.
• Companion scripts in scripts/ provide reference implementations for BIDS validation, phenotype extraction, and QC.

Core workflow (never bypassed):

1. Identify input REST-meta-MDD data and target modalities.
2. Generate a numbered execution plan clearly stating WHAT needs to be done and which tool skill will handle each step.
3. Present the full plan, estimated runtime, resource requirements, and risks to the user and wait for explicit confirmation ("YES" / "execute" / "proceed").
4. On confirmation, delegate every step to the appropriate skill via claw-shell.
5. After execution, save all outputs in a clean directory structure (rest_mdd_output/).

Research use only.

---

Quick Reference

Task	What needs to be done	Delegate to	Expected output
BIDS validation	Validate REST-meta-MDD BIDS structure	scripts/validate_rest_mdd.py	Validation report
rs-fMRI processing	Preprocessing, denoising, connectivity	fmri-skill	fmri_output/ connectivity
Phenotype extraction	Diagnosis, clinical measures, site info	scripts/extract_rest_mdd_phenotype.py	Merged phenotype CSV
Site harmonization	Multi-site effect correction	scripts/harmonize_sites.py	Harmonized data
QC summary	Per-subject quality control	scripts/rest_mdd_qc_summary.py	QC summary + exclusion list

---

Dataset Characteristics

• Cohort: ~3,600+ participants
  • MDD patients: ~1,837 Major Depressive Disorder patients
  • Healthy controls: ~1,779 age/sex-matched controls
• Sites: 17 research sites across China
• Scanner: Multi-site (various 3T scanners)
• Modalities: rs-fMRI (primary), T1w sMRI (some sites)
• Clinical: Diagnosis (SCID), HAMD, HAMA, medication status
• Access: Chinese Data Sharing Platform, REST-meta-MDD consortium
• Format: NIfTI (community BIDS conversion available)
• Reference: Yan et al. (2019), Science Bulletin

---

Supported Modalities

Modality	Description	Details
rs-fMRI	Resting-state functional MRI	Eyes closed, 5-10 min
T1w	Structural MRI (some sites)	1mm isotropic

---

REST-meta-MDD Clinical Measures

Measure	Description	Domain
Diagnosis	MDD vs. Healthy Control (SCID-based)	Clinical status
HAMD	Hamilton Depression Rating Scale	Depression severity
HAMA	Hamilton Anxiety Rating Scale	Anxiety severity
Medication	Medication status (medicated vs. drug-naive)	Treatment
Site	Data collection site (1-17)	Multi-site
Age	Age at scan	Demographics
Sex	Biological sex	Demographics
Education	Years of education	Demographics

---

BIDS Preparation

Script: scripts/validate_rest_mdd.py

Validates REST-meta-MDD BIDS structure and generates a compliance report.

python skills/rest-mneta-mdd-skill/scripts/validate_rest_mdd.py \
  --input /path/to/REST-meta-MDD/bids \
  --output /path/to/rest_mdd_output/qc/bids_validation.csv

Features:

• BIDS directory structure validation
• Site identification and completeness check
• Diagnostic group labeling (MDD vs. control)
• Modality completeness (rs-fMRI required, T1w optional)

---

Core Workflow (Never Bypassed)

1. Identify user target: full REST-meta-MDD processing, rs-fMRI only, phenotype extraction, or BIDS validation only.
2. Generate a numbered plan with tools, outputs, runtime, storage, and risks.
3. Wait for explicit confirmation (YES / execute / proceed).
4. On confirmation, run BIDS validation using scripts/validate_rest_mdd.py.
5. Delegate to fmri-skill for rs-fMRI processing.
6. If T1w data available, delegate to smri-skill.
7. If phenotype extraction is requested, run scripts/extract_rest_mdd_phenotype.py.
8. If site harmonization is requested, run scripts/harmonize_sites.py.
9. If QC summary is requested, run scripts/rest_mdd_qc_summary.py.
10. Save outputs into rest_mdd_output/.

---

Modality Processing Delegation

Modality	Delegated skill	Typical tasks	Main outputs
rs-fMRI	fmri-skill	preprocessing, denoising, connectivity	fmri_output/ connectivity
sMRI (T1w)	smri-skill	brain extraction, tissue segmentation	smri_output/ derivatives

---

Standard Output Layout

rest_mdd_output/
├── bids/                   # BIDS-staged data (or validation report)
├── fmri/                   # Functional MRI derivatives (rs-fMRI connectivity)
├── smri/                   # Structural MRI derivatives (if available)
├── phenotype/              # Merged phenotype tables (diagnosis, clinical, site)
├── harmonized/             # Site-harmonized data (ComBat or similar)
├── qc/                     # QC summaries and exclusion lists
└── logs/                   # Processing logs

---

Multi-Site Harmonization

REST-meta-MDD is a multi-site dataset (17 sites). Site effects are a major confound:

• ComBat: Commonly used batch effect correction for neuroimaging data
• Site-wise z-scoring: Normalize metrics within site before pooling
• Mixed-effects models: Include site as random effect in statistical analyses
• The scripts/harmonize_sites.py script provides reference implementations

---

Benchmark Adapter Guidance

For benchmark-style prompts, do not force the full orchestration when the task only asks for local REST-meta-MDD data validation.

• If the task starts from REST-meta-MDD data already present on disk and only asks for BIDS validation:
  • Skip the download stage
  • Default to the narrow path local REST-meta-MDD discovery -> BIDS validation -> report
• In benchmark mode, do not require explicit confirmation before presenting the validation solution.

---

Safety and Execution Policy

• No execution before explicit plan confirmation.
• All execution must be routed via claw-shell.
• Missing dependencies must be resolved by dependency-planner before running.

---

Important Notes and Limitations

• Multi-site data (17 sites) requires careful site effect handling.
• Scanner heterogeneity across sites introduces variability.
• rs-fMRI is the primary modality; structural data is limited.
• MDD diagnosis is SCID-based across all sites.
• Large sample size (~3,600) provides good statistical power for case-control analyses.
• Medication status is an important confound; subgroup analyses (medicated vs. drug-naive) are recommended.
• rest-mneta-mdd-skill is orchestration-only; detailed preprocessing logic remains in modality skills.

---

When to Call This Skill

• User asks for end-to-end REST-meta-MDD workflow.
• User asks to process REST-meta-MDD resting-state fMRI data.
• User needs BIDS validation for REST-meta-MDD data.
• User asks to extract REST-meta-MDD phenotype data (diagnosis, HAMD, site).
• User asks for depression neuroimaging analysis or multi-site harmonization.

---

Complementary / Related Skills

• fmri-skill → functional MRI preprocessing and analysis
• smri-skill → structural MRI preprocessing (if available)
• bids-organizer → BIDS validation and organization
• brain-visualization → visualization of derivatives
• dependency-planner → dependency resolution
• conda-env-manager → environment management
• claw-shell → command execution

---

Reference

• REST-meta-MDD: Chinese Data Sharing Platform
• Yan et al. (2019): Reduced default mode network functional connectivity in patients with recurrent major depressive disorder. Science Bulletin.
• REST-meta-MDD consortium

Created At: 2026-05-06 13:55 HKT Last Updated At: 2026-05-06 13:55 HKT Author: chengwang96