File Manifest for the S-BSST3199 Magnaporthe Colony Image Dataset

Provides a machine-readable file and image manifest for the research data deposited in EMBL-EBI BioStudies under accession S-BSST3199 (a time-series petri-dish image dataset of Magnaporthe colonies from twelve plates, with associated morphometric analysis outputs produced by metrics-petri 3.0.0). The original research files are not bundled in this R package; they remain hosted by BioStudies. The manifest can be used in image-analysis and plant-pathology workflows, including workflows based on the 'grayleafspotr' software. Related research outputs are documented using their persistent identifiers.

grayleafspotdata

A machine-readable manifest for the BioStudies record S-BSST3199, the time-series Magnaporthe colony image dataset deposited at EMBL-EBI. The package is a small data catalogue: it ships one data frame describing what is in S-BSST3199 and where each file lives, but it does not ship the 49 source images themselves.

Overview

This is a stand-alone R data package. It is not a fork of, and does not depend on, the grayleafspotr analysis software. The two packages can be used together, but no R-level dependency exists in either direction. The relationship is documentation-only.

The BioStudies record S-BSST3199 contains:

• 49 original petri-dish images of Magnaporthe colonies from 12 plates
• 12 plate-level image_metadata.csv files
• 12 extracted analysis output directories produced by metrics-petri 3.0.0
• The metrics-petri 3.0.0 source archive (archived separately at Harvard Dataverse, DOI 10.7910/DVN/SR2HBR)

Two acquisition groups are present. The 30 January group has 5 images of plate P001. The 6 February group has 44 images across plates P002 through P012, four per plate. Total: 49 source images.

Primary dataset

What this package contains

• grayleafspot_images: a data frame with 49 rows, one per source image. Columns include filename, relative_path, acquisition_group, plate_id, metadata_file, analysis_directory, study_accession, study_url, and provenance notes.
• grayleafspot_files: a data frame with 6 rows, one per top-level deposited file in S-BSST3199. Each row carries a SHA-256 checksum (from the deposited SHA256SUMS.txt), a file size (from the BioStudies files API), and a direct BioStudies file URL.
• inst/extdata/related_works.json: machine-readable metadata for the three persistent records linked from this package.
• inst/extdata/dataset_metadata.json: a small JSON sidecar summarising the primary dataset and compatible software.
• inst/extdata/build_summary.json: a small JSON report of the last manifest generation run (row counts, coverage, duplicates).
• inst/CITATION: an R CITATION file distinguishing four separately-citable resources.
• CITATION.cff: a CFF 1.2.0 citation file for repository hosting services. The CRAN source package excludes this file because R uses inst/CITATION for package citations.
• data-raw/build_manifest.R, data-raw/fetch_related_metadata.R, data-raw/validate_biostudies.R: developer scripts for regenerating the .rda files and the JSON sidecars from authoritative sources.

What this package does not contain

The 49 source images. The package does not embed raw_images.zip (≈ 146 MB), does not embed metrics-petri-3.0.0.zip (≈ 24 MB), and does not embed the individual image_metadata.csv files. The original files remain hosted by BioStudies. If you need a copy, fetch them from the canonical record:

https://www.ebi.ac.uk/biostudies/studies/S-BSST3199

The package also does not depend on grayleafspotr. Users may install grayleafspotr separately if they want to run analysis workflows on the downloaded images. No Depends, Imports, Suggests, Remotes, or LinkingTo field references it.

Data access

The manifest is installed when you install this package. The images themselves live at BioStudies. To download them, use either the BioStudies web interface or the documented file URL pattern:

https://www.ebi.ac.uk/biostudies/files/S-BSST3199/<flattened_filename>

For example, the top-level dataset_manifest.csv is reachable at:

https://www.ebi.ac.uk/biostudies/files/S-BSST3199/dataset_manifest.csv

The 49 individual source images are packaged inside raw_images.zip, so they do not have per-image direct URLs in the current BioStudies layout. To get one, download and extract raw_images.zip.

Dataset object

Two exported datasets are available after the package loads:

• grayleafspot_images: 49 rows, 18 columns.
• grayleafspot_files: 6 rows, 11 columns.

Column definitions and provenance notes for each are in the roxygen documentation (?grayleafspot_images, ?grayleafspot_files).

Example

library(grayleafspotdata)

data("grayleafspot_images")
data("grayleafspot_files")

head(grayleafspot_images)
nrow(grayleafspot_images)
# 49

unique(grayleafspot_images$study_accession)
# "S-BSST3199"

table(grayleafspot_images$acquisition_group, grayleafspot_images$plate_id)

# Files with verified SHA-256 (the six top-level deposited files)
grayleafspot_files[, c("filename", "file_size_bytes", "sha256")]

If you also install grayleafspotr separately, the manifest can guide file discovery for downstream analysis. No code in this package calls grayleafspotr.

Integrity and checksums

Each row in grayleafspot_files carries a SHA-256 value transcribed directly from the deposited SHA256SUMS.txt. They are the only checksums the BioStudies record publishes at this level.

Per-image SHA-256 values in grayleafspot_images are NA in the published build, because the BioStudies record only checksums the top-level raw_images.zip rather than its contents. To populate per-image checksums, point data-raw/build_manifest.R at a local copy of the dataset:

Sys.setenv(GRAYLEAFSPOT_DATA_DIR = "/path/to/local/S-BSST3199")
source("data-raw/build_manifest.R")

The script will compute SHA-256 with openssl::sha256() (falling back to digest::digest(..., algo = "sha256")) and join them onto the manifest. No personal absolute paths are stored in the published .rda. The local_manifest_source column records which directory was used during regeneration; it is NA in the remote-only build.

Compatible analysis software

This data package is independent of grayleafspotr. The manifest and underlying research files may be used as inputs to workflows based on grayleafspotr, including builds distributed through the relevant R-universe repositories. No dependency between this data package and grayleafspotr is imposed.

Reference: https://rotsl.r-universe.dev/builds

If there are two separately displayed builds or sources for grayleafspotr (for example, the Bioconductor-staged build at BiocStaging/grayleafspotr), those are software distribution references. They are not separate copies of the research dataset.

Related research outputs

Each row is a distinct citable resource with its own licence, author list, and publication date. S-BSST3199 is the canonical record for the biological images and their derived outputs. The two Dataverse records are related but separate: SR2HBR archives the analysis software, and 7BJLIQ archives the U-Net model trained on 605 images.

Citation

If you use this R manifest package, cite the package itself. If you use the underlying biological images or analysis outputs, cite S-BSST3199 according to the BioStudies record. If your analysis relies on the archived software or the U-Net validation dataset, cite the corresponding Harvard Dataverse DOI as well. Do not merge these into a single false citation.

Citation details are in inst/CITATION (R citEntry format). The repository root also contains CITATION.cff (CFF 1.2.0) for repository hosting services, but that file is not part of the CRAN source package.

library(grayleafspotdata)
citation("grayleafspotdata")

Data provenance

The manifest is generated by data-raw/build_manifest.R from the authoritative BioStudies record. Inputs to the build:

• https://www.ebi.ac.uk/biostudies/files/S-BSST3199/dataset_manifest.csv (49 image rows)
• https://www.ebi.ac.uk/biostudies/files/S-BSST3199/SHA256SUMS.txt (6 top-level checksums)
• https://www.ebi.ac.uk/biostudies/api/v1/studies/S-BSST3199/files (top-level file sizes)

Each row's metadata_source column records where its values came from. The build summary at inst/extdata/build_summary.json records the build date, row counts, and integrity flags.

Source-data provenance (the BioStudies record itself) is distinct from R-package provenance (this repository). The first says where the biological data came from. The second says how the manifest was assembled.

Reproducibility

The manifest regenerates deterministically. build_manifest.R sorts rows by relative_path then filename before saving, so two runs against the same inputs produce byte-identical .rda files (modulo the .rda compression wrapper, which is itself deterministic).

To rebuild from scratch:

Sys.setenv(GRAYLEAFSPOT_DATA_DIR = "/path/to/local/S-BSST3199")
source("data-raw/fetch_related_metadata.R")  # refresh JSON sidecars
source("data-raw/build_manifest.R")          # refresh .rda files
source("data-raw/validate_biostudies.R")     # write validation report

To re-run package checks:

devtools::document()
devtools::test()
devtools::check()

License

The R package source, the manifest-generation scripts, and the machine-readable metadata files in this repository are licensed under the MIT license. See LICENSE and LICENSE.note.

This does not change the licensing terms of the underlying research data in BioStudies S-BSST3199, or of the related Harvard Dataverse records. Those records are licensed independently and remain governed by their own deposit terms (both Dataverse records are MIT-licensed; the BioStudies record is governed by its own deposit terms).

Contributing / corrections

Corrections to the manifest are welcome. If a row is wrong, the fix is to regenerate the manifest with the corrected input, not to edit the .rda file by hand. Open an issue at https://github.com/rotsl/grayleafspotdata/issues describing what is wrong and how to reproduce it.

Changes that affect either grayleafspotr package are out of scope here. Those packages are not modified by, or as a consequence of, work on this repository.