Meisler et al. ABCC dMRI Project

Code, analysis workflows, and reproducibility documentation for the Meisler et al. ABCC diffusion MRI paper.

This repository covers:

• dMRI post-processing and tabular data assembly
• harmonization across scanner/device-software batches
• modeling of batch, age, and quality effects
• main-text and supplemental figure generation
• notebook-based manuscript results summaries

Start Here

The fastest way to get oriented is the reproducibility guide https://pennlinc.github.io/Meisler_ABCD_dMRI/:

• Reproducibility Guide index
• Replication setup
• Modeling workflow
• Main figures
• Supplemental figures

If you are replicating the paper analyses (rather than reprocessing raw imaging), begin at Chapter 6 and proceed in order.

Repository Structure

.
├── config.json                  # Machine-specific paths + shared analysis settings
├── data/                        # Derived tabular data inputs/outputs
├── figures/                     # Generated figure outputs
├── reproducibility_guide/       # Jupyter Book source for full methods/replication docs
└── scripts/
    ├── 0_processing/            # QSIPrep/QSIRecon processing scripts and configs (not to be replicated)
    ├── 1_munge_data/            # Build and filter merged analysis table
    ├── 2_harmonize_data/        # Longitudinal ComBat-GAM harmonization
    ├── 3_batch_effects/         # Batch effects modeling
    ├── 4_age_effects/           # Age effects modeling (pooled + vendorwise)
    ├── 5_quality_effects/       # IQM/manual quality effects modeling
    ├── 6_figures/               # Main-text figure notebooks + tract plotting scripts
    ├── 7_results/               # Notebook-generated manuscript/table statistics
    └── 8_supplement/            # Supplemental figure notebooks and table scripts

Quick Setup

1. Set up config.json for your machine (especially project_root, r_env, python_env).
2. Export CONFIG_PATH so scripts can find your config.
3. Install required R/Python packages (see Replication setup).

export CONFIG_PATH="/absolute/path/to/this/repo/config.json"

Analysis Flow

1. Data munging/filtering (scripts/1_munge_data)
2. Harmonization (scripts/2_harmonize_data)
3. Modeling (scripts/3_batch_effects, scripts/4_age_effects, scripts/5_quality_effects)
4. Main figures/results (scripts/6_figures, scripts/7_results)
5. Supplemental analyses (scripts/8_supplement)

Detailed commands and assumptions are documented in the reproducibility guide chapters.

Note some steps are PennLINC-internal or depend on ABCC/ABCD controlled-access infrastructure.