Skip to content

spec.md

Latest commit

 

History

History
55 lines (41 loc) · 3.37 KB

spec.md

File metadata and controls

55 lines (41 loc) · 3.37 KB
Raw

systemPipeShiny — Product Spec

Precedence: spec.md > sub-spec > Plans.md. This file defines what is correct; Plans.md tracks what to do.

Purpose

systemPipeShiny (SPS) is a Bioconductor R package that provides a Shiny GUI on top of systemPipeR (SPR). Users who don't know R can create, configure, and run SPR workflows, visualize the results, and build their own Shiny tabs without writing Shiny code.

Distribution

  • Released through Bioconductor. The canonical source is https://git.bioconductor.org/packages/systemPipeShiny (git remote upstream); the GitHub mirror is systemPipeR/systemPipeShiny (git remote origin).
  • Branch devel tracks Bioconductor devel. Bioconductor's build system pushes version bumps (x.y.z, with even y for release and odd y for devel), so pull from upstream before starting work.
  • The package must pass R CMD build and R CMD check (including BiocCheck) with no ERRORs or WARNINGs.
  • AI, agent, and planning files must never be part of the built tarball (see .Rbuildignore).

Public API (contract)

The exported functions in NAMESPACE are the public contract. Changing a function's signature or behavior counts as a breaking change and needs a NEWS.md entry.

Area Exports
App lifecycle spsInit() creates an app project from inst/app; sps() builds the UI/server
Options spsOptions(), spsOptDefaults()
Tabs spsNewTab(), spsEzUI(), spsEzServer(), spsTabInfo(), removeSpsTab(), spsCoreTabReplace()
Data/UI helpers dynamicFile(), dynamicFileServer(), loadDF(), genGallery(), genHrefTable(), canvasBtn()
Accounts / DB (R6) spsDb, spsEncryption, spsAccount (SQLite-backed, openssl encryption)

Functional modules

  1. Workflow: target/metadata editor, workflow step selection, CWL editing, running workflows via systemPipeR.
  2. Visualization (RNA-seq): normalization, DEG (DESeq2), and plots (PCA, t-SNE, GLM-PCA, heatmap, UpSet, dendrogram).
  3. Quick ggplot: esquisse-based plotting.
  4. Canvas: collect, arrange, and export plots across tabs.
  5. Admin and login: optional user/admin pages, backed by spsAccount.
  6. Custom tabs: templates in inst/app/templates, registered through config/tabs.csv.

Invariants

  • Heavy and module-specific packages (systemPipeR, DESeq2, esquisse, …) stay in Suggests, and modules check for them at runtime (checkModulePkgs_internal). They must not move into Imports.
  • The app project scaffold (inst/app) and config/sps_options.yaml must stay in sync with spsOptDefaults().
  • Tests under tests/testthat must skip, not fail, when the session is non-interactive and an interactive dependency is missing.
  • Credentials and DB keys are never committed. The SQLite DB is created at spsInit() time.
  • R code in R/ never references raw files on external hosting platforms (raw.githubusercontent.com, GitLab, Bitbucket, Dropbox). BiocCheck treats these as an ERROR. Ship the asset in inst/ instead, or read the installed copy (e.g. system.file("NEWS.md", package = "systemPipeShiny")).

Quality gates

  • R CMD build . && R CMD check --as-cran systemPipeShiny_*.tar.gz passes with 0 ERRORs and 0 WARNINGs.
  • BiocCheck::BiocCheck() reports no ERRORs.
  • testthat passes.
  • No lint/formatter baseline is configured yet. Any plan that changes R code should first add a lintr setup task.