PRISM Studio (Web Interface)

PRISM Studio (prism-studio.py) is the web interface for validating datasets and interactively working with metadata.

1) Start PRISM Studio

macOS / Linux:

source .venv/bin/activate
python prism-studio.py

Windows:

.venv\Scripts\activate
python prism-studio.py

Open:

  • http://localhost:5001

Quick health checks (repo root):

# fast entrypoint sanity
bash scripts/ci/run_local_smoke.sh

# full required gate
bash scripts/ci/run_runtime_gate.sh

2) Upload and validation workflow

What you can upload

  • Folder (recommended)

  • ZIP file

Large datasets (“structure-only uploads”)

PRISM Studio supports a “DataLad-style” upload strategy for large datasets:

  • Metadata files (e.g., .json, .tsv) are uploaded.

  • Large binaries (e.g., .nii, .mp4) may be skipped.

  • The backend creates placeholders so PRISM can validate structure, filenames, and sidecar/schema consistency without transferring large payloads.

This is useful when you only need to validate correctness of the dataset layout and metadata.

3) Understanding the results

After validation, PRISM Studio displays:

  • A summary of Errors (must fix) and Warnings (should fix)

  • Per-issue details including which file is affected and why it failed

Typical errors include:

  • Missing required BIDS files (e.g., dataset_description.json)

  • Missing sidecar JSON for a TSV/recording

  • Filename not matching expected patterns

  • JSON fields failing schema validation

You can typically download a JSON report for archiving or CI workflows.

4) JSON editor

PRISM Studio includes a JSON editor that can:

  • Create/edit sidecars (*_beh.json, *_physio.json, biometrics JSON sidecars)

  • Validate JSON content against the selected schema version

This helps when building metadata iteratively.

5) Template Editor (Survey / Biometrics)

PRISM Studio includes a Template Editor to create or modify PRISM library templates (Survey or Biometrics) without editing raw JSON.

What it supports:

  • Load a template from a library folder or create a new template from the selected PRISM schema version

  • Value-only editing: keys stay fixed/locked; the UI renders fields as form controls

  • No brackets for typical fields: arrays/levels/translations are edited via add/remove controls and language rows

  • Schema help: hover the ⓘ icons to see descriptions and allowed values

  • Validate against the selected schema and download the JSON (no server-side overwrite)

Where to find it:

  • From the navbar: Template Editor

  • Or open directly: http://localhost:5001/template-editor

Tips:

  • Use Add in the left “Items (questions/metrics)” panel to add a new item ID (it becomes fixed once created).

  • For translated text fields, language codes (e.g., de, en) are non-editable; only the text values are editable.

6) NeuroBagel integration (participants)

PRISM Studio provides a participants annotation workflow compatible with NeuroBagel:

  • Reads participants.tsv

  • Helps you annotate with standardized ontology terms

  • Supports categorical level extraction

7) Methods Boilerplate Generator

PRISM Studio can generate a draft of a scientific Methods section based on the metadata in your library.

  • How it works: It scans your library/survey and library/biometrics folders.

  • Output: A Markdown document describing the instruments used, their authors, citations, and constructs.

  • Languages: Supports generating text in English and German.

  • Metadata: The generated text now pulls canonical DOIs, licenses, target age ranges, estimated administration/scoring times, item counts, and access classifications from the tightened PRISM schema so each instrument summary reflects the richer metadata you provided.

Access it via the Tools menu or at /methods-generator.

8) Survey Generator & Export

The Survey Export tool allows you to:

  • Select items from the library.

  • Preview response scales (levels), units, and value ranges.

  • Export a “clean” dataset structure for data collection.

  • Generate codebooks automatically.

8) Tips for first-time use

  • Start with a small demo dataset: see WORKSHOP.md.

  • Validate first, then iterate on metadata fields.

  • If you want strict automation, use prism.py in your terminal.