Uploading Datasets

This guide walks you through uploading a BIDS dataset to NEMAR.

Prerequisites

Before uploading:

• Dataset is in valid BIDS format
• Logged in with nemar auth login
• git-annex installed
• GitHub CLI (gh) installed and authenticated
• Sandbox training completed (nemar sandbox)

DataLad Users

NEMAR requires main as the default branch. DataLad’s adjusted branch naming (e.g. adjusted/master(unlocked)) will cause CI and metadata pipelines to fail.

The CLI will detect non-main branches during upload and offer to rename automatically. You can also rename manually:

git branch -m adjusted/master\(unlocked\) main

Step 1: Validate Your Dataset

Always validate before uploading:

nemar dataset validate ./my-dataset

Validation Must Pass

Datasets with validation errors cannot be uploaded. Fix all errors before proceeding.

Common Validation Issues

Issue	Solution
Missing dataset_description.json	Create the required BIDS metadata file
Invalid JSON	Check for syntax errors in JSON files
Missing required fields	Add Name and BIDSVersion to dataset_description.json
Invalid modality data	Ensure data files match BIDS naming conventions

Step 2: Upload

nemar dataset upload ./my-dataset

Options

Option	Description
--name, -n	Dataset name (defaults to BIDS Name field, then directory name)
--description	Brief description
--skip-validation	Skip BIDS validation (not recommended)
--skip-orcid	Skip co-author ORCID collection
--dry-run	Show what would be uploaded without doing it
--restart	Clear upload progress and re-upload all files
-j, --jobs	Number of parallel upload jobs (default: 4)
-y, --yes	Skip confirmation and proceed

Step 3: What Happens

The upload process:

1. Prerequisite Check - Verifies required tools (git-annex, gh, aws) are installed with platform-specific install guidance if missing
2. Auth and Prerequisites - Verifies login, GitHub authentication (HTTPS preferred, SSH as fallback)
3. BIDS Validation - Runs the official BIDS validator (unless skipped)
4. File Manifest - Collects files and co-author ORCIDs
5. License Enforcement - Detects license from dataset_description.json or LICENSE file; prompts to select one if missing. Validates the license allows research redistribution (see License Requirements below)
6. Data Provenance - For derived datasets, collects source dataset DOIs and checks license compatibility
7. Confirmation - Shows upload plan for review
8. Dataset Registration - Creates dataset record and private GitHub repo
9. GitHub Invitation - Accepts collaborator invitation to the repo
10. git-annex Init - Initializes git-annex and configures S3 remote
11. Data Upload - Uploads large files to S3 (uses AWS CLI fast-path when available)
12. Metadata and Push - Writes metadata, commits, and pushes to GitHub
13. CI Deployment - Deploys GitHub Actions workflows for validation

License Requirements

Every dataset uploaded to NEMAR must have a license that allows research redistribution. The CLI will:

1. Auto-detect your license from dataset_description.json (the License field) or a LICENSE file
2. Prompt you to choose if no license is found, offering recommended open data licenses
3. Validate that the chosen license allows research use
4. Create a LICENSE file if one does not exist

Recommended licenses (most permissive to most restrictive):

License	Description
CC0-1.0	Public domain dedication (no restrictions)
PDDL-1.0	Public Domain Dedication and License
CC-BY-4.0	Attribution only
CC-BY-SA-4.0	Attribution + ShareAlike
CC-BY-NC-4.0	Attribution + NonCommercial
CC-BY-NC-SA-4.0	Attribution + NonCommercial + ShareAlike
ODC-By-1.0	Open Data Commons Attribution
ODbL-1.0	Open Database License

You can also enter any valid SPDX license identifier manually if it allows research use.

License for Derived Data

If your dataset is derived from another dataset, the CLI will ask about data provenance and check that your chosen license is compatible with the source dataset’s license. For example, a dataset derived from CC-BY-SA-4.0 source data must also use CC-BY-SA-4.0.

GitHub Authentication

The CLI uses HTTPS-first authentication for GitHub operations:

1. CI/CD: Uses GH_TOKEN environment variable if set
2. Local (preferred): Uses gh auth token from GitHub CLI, so run gh auth login first
3. Fallback: SSH key authentication (nemar auth setup-ssh)

Step 4: Making Updates

After initial upload, push changes using the CLI:

cd nm000104  # Your dataset directory

# Make changes, then save and push
nemar dataset save -m "Add subjects 101-110"
nemar dataset push

Or create a formal update PR:

nemar dataset update ./nm000104

Troubleshooting

Upload Fails with Authentication Error

# Check login status
nemar auth status --refresh

# Re-login if needed
nemar auth login

git-annex Errors

# Ensure git-annex is configured
git annex version

# Re-initialize if needed
git annex init

Upload Interrupted or Timed Out

The upload tracks progress automatically. Re-run the same command to resume:

# Resume from where it left off
nemar dataset upload ./my-dataset

# Or start fresh if resume fails
nemar dataset upload ./my-dataset --restart