What problem does it solve? Processing the REST-meta-MDD multi-site depression neuroimaging dataset requires coordinating BIDS validation, resting-state fMRI preprocessing, clinical phenotype extraction, site harmonization, and quality control across roughly 3,600 subjects from 17 sites, which is error-prone when done manually. ## Core Features & Use Cases - BIDS Validation: Scans REST-meta-MDD subject directories, checks for required rs-fMRI files, detects site IDs, and produces a compliance report CSV. - Phenotype Extraction: Merges diagnosis, HAMD, HAMA, medication, and demographic columns from heterogeneous phenotype files into one subject-aligned table. - Multi-Site Harmonization: Applies ComBat or site-wise z-score normalization to correct scanner and site effects across the 17 collection sites. - QC Summaries: Computes framewise displacement metrics from fMRIPrep confounds and generates per-subject exclusion lists with site-level summaries. - Use Case: A researcher downloads REST-meta-MDD data and asks for an end-to-end pipeline; the skill validates the BIDS structure, delegates rs-fMRI preprocessing to fmri-skill, extracts clinical phenotypes, harmonizes site effects, and outputs a clean rest_mdd_output/ directory. ## Quick Start Ask the agent to run the full REST-meta-MDD pipeline on your local BIDS dataset, starting with validation and proceeding through rs-fMRI processing, phenotype extraction, and QC after you confirm the plan.
