Skip to contents

The metadata Shiny app

Source: vignettes/metadata-app.Rmd
metadata-app.Rmd

Overview

A TRIA submission needs metadata at every level of the sampling hierarchy – site, tree, woodpiece, slide – plus dataset-level information such as authors, funding, licensing, and related publications (see vignette("QWAmetadata") for the full list of components). Filling all of this in by hand, as data frames or JSON, is tedious and error-prone.

The metadata Shiny app is a guided, form-based alternative: it takes the QWAimages object you already built (Step 3 of vignette("workflow")) as its starting point and walks you through the remaining components tab by tab. Along the way it offers lookup tools for organisations (ROR), authors (ORCID) and related publications/datasets (Crossref DOIs), and validates your entries as you go. The output is a QWAmetadata object, exported as a .json file – one of the three files required for every TRIA submission.

The app is entirely self-contained: it does not read or write anything outside of what you explicitly load or export, and any partially completed submission can be saved and reopened later.

Launching the app

This opens the app in your system’s default web browser (or the RStudio/Positron viewer pane). It requires a number of Suggested packages (bslib, data.tree, DT, htmlwidgets, httr, leaflet, rcrossref, rhandsontable, sf, shiny, shinyjs, shinyTree, shinyvalidate) which are only needed for this app – rxs2tria will prompt you to install any that are missing.

(screenshot: app opened on the Start tab)

Walking through the tabs

The app has four tabs, meant to be completed in order (a “Next”/“Previous” button pair in each sidebar moves between them), though you can jump back to fix earlier entries at any time.

1. Start

Load your input data using the Load input data button in the sidebar, from one of three sources:

  • From local R environment: point it at a QWAimages or QWAmetadata object already in your R session (e.g. rxs_images from Step 3 of vignette("workflow")).
  • From file: a QWAimages .csv file, or a QWAmetadata .json file – use the latter to resume a partially completed submission.
  • Load example data: loads the small example dataset bundled with the package, useful for trying out the app without your own data.

(screenshot: input source modal)

Once loaded, the main panel shows:

  • Data structure: an interactive tree (site > tree > woodpiece > slide > image) visualising the hierarchy inferred from your image file names, with the labels rxs2tria generated for each level.
  • Metadata extracted from files: a table of everything extracted from the ROXAS/ROXAS AI settings and image files. Most columns are read-only; only_ew and img_comment can be edited directly in the table (which supports Excel-like copy/paste and autofill).

(screenshot: data structure tree and metadata table)

Check this tab carefully – it is the foundation for everything that follows. If the structure or labels look wrong, revisit Step 2 (extract_data_structure()) in vignette("workflow") and start over.

2. General

Dataset- and author-level information, in three parts:

  • Dataset information: title, description, access rights (public or restricted, with an embargo date), licence (CC licences preferred), and acknowledgements.
  • Authors & Funding: two editable tables for authors and their affiliations as well as funding sources. Two search tools for ROR and ORCID data can help you find the relevant information.
  • Related resources: a table of related publications or datasets, with a Crossref-backed search by DOI, citation text, or TRIA dataset ID.

(screenshot: author table with ORCID search results)

A validation check summary in the sidebar lists any issues with the entries so far (e.g. missing required fields).

3. Samples

Site-, tree-, woodpiece- and slide-level metadata, in one table per level. Rows are pre-filled with the IDs and labels inferred from the QWAimages or (if available) taken directly from the QWAmetadata provided as input in the Start tab. The site table also drives an interactive map: any site with valid coordinates gets a marker, which is a quick way to spot mistyped coordinates.

(screenshot: site map with markers)

4. Summary

A combined validation check summary across all tabs – resolve (or consciously accept) any remaining issues here. Clicking Export data to file downloads the completed QWAmetadata object as a .json file.

(screenshot: summary tab with validation table)

Output and how it fits into the workflow

The exported .json file is a QWAmetadata object – see vignette("QWAmetadata") for its structure and for how to inspect, check, or edit it further in R. Read it back with:

QWA_meta <- read_QWAmetadata("path/to/output_data/example_dataset_QWAmetadata.json")

Within the overall data preparation workflow (vignette("workflow")), this app covers Step 3b, run after building the QWAimages object in Step 3 and before (or alongside) processing the measurement data in Steps 4-5. The resulting .json file is, together with the two QWAdata .csv files, one of the three files required for every TRIA submission (vignette("submission")).

Because partially completed submissions can be saved and reopened at any time, you do not need to finish in one sitting – export what you have, and load the .json back into the Start tab later to continue.