Repository navigation
Conversation
…flash - Default a freshly uploaded document to checked for the next question, and switch its checkbox id from row index to sanitized name so a doc's input state stays attached to it when uploads/deletes reorder the list. - mark_consumed() now only writes the consumed reactiveVal when the locked set actually changes, instead of on every ask - a reactiveVal invalidates its renderUI dependents on every write regardless of whether the value changed, which was causing the whole Sources docs list to visibly flash on every single question, not just the first one that locks a document. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RjAHhKH71SHFbo2KuMcBjE
fix(obi): warn instead of silently blocking uncheck of documents already in chat context
- swap the "multi-omics median" choice for "multiomics", since normalizeMultiOmics() dropped that algorithm on 2025-11-11 and has ignored the "median" value ever since - translate a legacy-recorded "median" through selectInput() so it resolves to the new choice instead of silently falling back to choices[[1]]
- run upload previews through the shared preprocessing pipeline - preserve literal source matrices across compute and reanalysis - align downstream views explicitly when processed dimensions differ
- repoint pp.* and pgx.preprocess calls from playbase to the new playbase.preprocess leaf package across upload, app_qsee, and correlation board code, keeping the numerical pipeline behind one dedicated boundary - register playbase.preprocess in DESCRIPTION, dev/functions.R, and dev/write_description.R so it is fetched and tracked like the other bigomics dependencies - update GSE/TCGA pipeline scripts and the upload-preprocess contract test to match the new call paths
- split the deployment licence gate (ENABLE_AI) from the provider pin (AI_PROVIDER_LOCKED): pinning the provider no longer freezes the AI on/off switch or the data-sharing consent for every non-admin - the free deployment sets AI_PROVIDER_LOCKED=TRUE purely to stop users picking a provider, which left all seven AI controls disabled and AI forced on with no way to opt out, despite the data-exposure warning - add regression tests for that deployment config and re-point the consent tests that encoded the old coupling at the licence gate
- pass batch.correct.method and batch.pars from the panel to pgx.createPGX - align legacy datasets by name so every board opens without an alignment - migrate qsee plot calls to the omicsplots control-bundle arguments
- use the bigomics/playbase.preprocess shorthand so access is authenticated - emit the shorthand from the generator instead of a url archive fetch
…eprocessing # Conflicts: # VERSION
Contributor
Author
|
Please @ivokwee confirm if we close this PR in favour of using the branch as reference for fanta. |
Contributor
Author
|
Implementation in shiny OPG cancelled. This will go directly to fanta |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
The app stops calling preprocessing primitives scattered across playbase and talks to one boundary instead. Two consequences shape this PR:
compute runs.
countsis the pure user input, so boards must ask for it aligned.pgx$countskeeps its uploaded shape and no longer matchespgx$X..opg_pgx_analysis_counts()is the only supported way to read counts on X's axes — never indexpgx$countsdirectly in a board.Merge order
Depends on bigomics/playbase#526
playbase→ this PR.Testing
tests/testthat/test-upload-preprocess-contract.R)example-data.pgx,mouse-example.pgx,arabidopsis.pgxand the AML multi-omics dataset
omicsplots::arguments