Processing & Derivative Data Standards🔗

All pipelines used for HBCD data processing must follow HBCD processing and derivative data standards, including: NMIND standards for reproducible neuroimaging and HBCD-specific requirements.

NMIND Checklists (Bronze Badge Items)🔗

The NMIND consortium is dedicated to the advancement of community standards and utilities in support of reproducible neuroimaging research (Kiar et al. 2023). For HBCD, peer review of software is performed via submission of the NMIND Coding Standards Checklist. Software is reviewed against community-driven scientific software standards across three main categories: documentation, infrastructure, and testability. Reviewed tools are assigned badge ratings and published to Evaluated Tools. At a minimum, HBCD pipelines must meet Bronze-level standards. Pipelines are additionally published with DOIs to support reproducibility.

NMIND Documentation Standards Checklist🔗

Landing page provides a link to documentation and brief description of what program does
Documentation is up to date with version of software (see Obtaining a DOI for details)
Typical intended usage is described
An example of its usage is shown
Document functions intended for users (i.e., public function docstring/help coverage ≥ 10%)
Reasonable description of required inputs (i.e., "NIfTI of brain mask in MNI" vs. "An image file")
Description of output(s)
User installation instructions available
Dependencies listed (Note: this is largely inapplicable to HBCD pipelines as they are required to be containerized - see details)

NMIND Infrastructure Standards Checklist🔗

Code is open source
Package is under version control (see Version Control for details)
Readme is present
License is present (see Licensing Tips for details)
Issues tracking is enabled (i.e., either through GitHub or external site)
Digital Object Identifier (DOI) points to latest version (see How to Obtain a DOI for details)
All documented installation instructions can be successfully followed

NMIND Testing Ability Standards Checklist🔗

Provide/generate/point to test data (Only applicable if non-HBCD data can be shared as a representative data sample)
Provide instructions for users to run tests and evaluate for correct behavior

HBCD-Specific Requirements🔗

Additional requirements include compliance with HBCD Study derivative data standards and Brain Imaging Data Structure (BIDS). Pipelines must also be implemented as BIDS Apps (Gorgolewski et al.,2017), ensuring containerized, standardized processing.

Derivative Data Standards🔗

All software standards also apply to derivative data standards for HBCD Study data. In practice, this means software must comply with the data standards in order for its processed outputs to be eligible for inclusion in a release. Therefore, in addition to the requirements outlined below, all pipelines must produce derivatives compatible with current release data.

BIDS Compliance & BIDS App Implementation🔗

BIDS is a community-driven standard for organizing neuroimaging and behavioral data to make datasets structured, shareable, and reproducible. BIDS Apps are containerized applications that intake and output BIDS-formatted data.

Per the BIDS-App guidelines, all HBCD pipelines must be containerized to run on any system supporting Docker or Apptainer (Singularity). Containerization ensures that all software dependencies are included to guarantee consistent processing environments across systems. This is not only critical for reproducibility, but also makes data processing more straightforward and foolproof.

Dedicated Pipeline Documentation Website🔗

HBCD pipelines must maintain living documentation via a dedicated website. This provides a current, easily navigable resource beyond static publications or a GitHub README. We recommend hosting documentation with Read the Docs or GitHub Pages. See the fMRIPrep documentation for guidance on organization of level of detail.