Preserve computed GUI runs as selectable sessions (#297)

ChangeLog

@@ -14,6 +14,12 @@
 
 ## Interface and handling
 
+- reorganize GUI parameters, add conditional seed/accessibility controls and
+  grouped CSV defaults, expand FASTA limits and refine result headers (PR #299)
+
+- retain computed GUI runs as named, foldable sessions with JSON input restoration,
+  independent outputs, per-session ZIP export and deletion (PR #297 review)
+
 - permit clearing inherited CSV sorting when replaying JSON as text reports;
   validate JSON/GUI behavior in native release/debug and real browsers (#287)
 
@@ -279,6 +285,40 @@ energies, restricted partition sums, and trackers.
 ################################################################################  
 ################################################################################
 
+261024 Alexander Mitrofanov
+ * gui/{app,options,columns}.js, gui/{index.html,styles.css}
+   * organize tabs/rows as Sequences, Seed, Output, Model & Energy, Query,
+     Target and CLI-only; align wrapped labels and place Compute first
+   + add a three-way seed selector, synchronized accessibility flags and
+     mode-dependent input availability; place SHAPE beside accessibility
+   * group compact CSV choices with descriptions and group selection actions;
+     default to 10 predictions and expanded task-specific energy columns,
+     explicitly preserving browser defaults in native-compatible JSON
+ * gui/{protocol,uploads,parameter-file}.js, gui/examples/
+   * accept 50 FASTA records per strand of 2-5,000 nt each, with longest
+     selected query x longest selected target <= 1,000,000; allow 1 MB FASTA
+   + warn about ignored FASTA header descriptions while using first-token IDs
+ * gui/{sessions,results}.js
+   * select results from the header/name, order Show/Hide before the name and
+     ZIP/delete actions, compact table headings and distinguish selected rows
+ * gui/tests/, scripts/wasm/test-browser.py, README.md, gui/docs/
+   + cover limits, native JSON replay/defaults, dependent controls, grouped CSV
+     choices, desktop/mobile alignment and result header selection
+   * address https://github.com/BackofenLab/IntaRNA/pull/299#issuecomment-6375752155
+
+261023 Alexander Mitrofanov
+ * gui/{app,sessions,results,visualization}.js, gui/{index.html,styles.css}
+   + retain successful runs in browser memory with unique IDs, editable names,
+     independent results/diagnostics and JSON snapshots of their computed inputs
+   * restore inputs and fold other output areas on selection; preserve sessions
+     across edits, reset, imports, failures and aborts; unload folded viewers
+   + export each session's original JSON/outputs from its header and delete it
+     without overwriting editable inputs; keep controls keyboard accessible
+ * scripts/wasm/{stage.sh,test-browser.py}, README.md, gui/docs/
+   * stage the sessions module, exercise session lifetime and archive isolation
+     in real browsers, and document temporary storage and updated reset behavior
+   * follow-up to https://github.com/BackofenLab/IntaRNA/pull/297#issuecomment-6367326708
+
 261023 Alexander Mitrofanov
  * .gitattributes, gui/docs/json-workflow.md
      + Preserve exact upstream license whitespace and mark the JSON header and

README.md

@@ -6,7 +6,9 @@
 seeding of interaction sites**
 
 Predict short RNA pairs or evaluate supplied structures locally with the [browser interface](https://backofenlab.github.io/IntaRNA/gui/).
-Select an interaction to view it in vaRRI, or download a ZIP with inputs, compiled versions, reproducible parameters and all outputs.
+The browser accepts up to 50 records per strand, 2–5,000 nucleotides each, with longest selected query × longest selected target at most 1,000,000. Prediction defaults to 10 interactions per pair and includes opening and seed energies; evaluation selects all energy columns.
+Computed runs stay in separate, foldable sessions under Results in the current tab. Name and select sessions to restore their inputs and compare results; editing or resetting parameters keeps earlier runs.
+Select an interaction to view it in vaRRI, or download a session’s ZIP with inputs, compiled versions, reproducible parameters and all outputs. Sessions are temporary and are cleared when the page is reloaded.
 See the [WebAssembly build instructions](gui/docs/step1-wasm-compilation.md) and
 [interface/deployment notes](gui/docs/step4-results-and-deployment.md).
 

gui/columns.js

@@ -313,3 +313,23 @@ export const CSV_CATALOG = [
     "numeric": true
   }
 ];
+
+// Every native column appears once. Ensemble energies and their components are
+// grouped with probabilities so users can see the cost/meaning of those outputs.
+export const CSV_GROUPS = [
+  { label: 'Energy', keys: ['E', 'ED1', 'ED2', 'E_init', 'E_loops', 'E_dangleL', 'E_dangleR', 'E_endL', 'E_endR', 'E_hybrid', 'E_norm', 'E_hybridNorm', 'E_add', 'seedE', 'seedED1', 'seedED2'] },
+  { label: 'Output', keys: ['subseqDP', 'subseqDB', 'hybridDP', 'hybridDB', 'hybridDPfull', 'hybridDBfull', 'bpList'] },
+  { label: 'Sequence', keys: ['id1', 'id2', 'seq1', 'seq2', 'subseq1', 'subseq2'] },
+  { label: 'Interaction', keys: ['start1', 'end1', 'start2', 'end2', 'seedStart1', 'seedEnd1', 'seedStart2', 'seedEnd2'] },
+  { label: 'Partition Function', keys: ['Pu1', 'Pu2', 'seedPu1', 'seedPu2', 'w', 'Etotal', 'Eall', 'Eall1', 'Eall2', 'Zall', 'Zall1', 'Zall2', 'EallTotal', 'P_E', 'RT'] },
+];
+
+export function defaultColumns(task) {
+  const context = ['id1', 'start1', 'end1', 'id2', 'start2', 'end2', 'subseqDP', 'hybridDP', 'E'];
+  // Seed energies remain available in evaluation, where native output is NAN.
+  // List them explicitly: seedEnd1/2 are coordinates, not energy columns.
+  const extra = task === 'evaluation'
+    ? [...CSV_CATALOG.filter(({ key }) => key.startsWith('E')).map(({ key }) => key), 'seedE', 'seedED1', 'seedED2']
+    : ['ED1', 'ED2', 'seedE', 'seedStart1', 'seedStart2'];
+  return [...new Set([...context, ...extra])];
+}

gui/docs/json-workflow.md

@@ -1,6 +1,7 @@
 # JSON settings and reproducible runs
 
-Load and Store sit to the right of Reset parameters. Store downloads
+The action bar starts with Compute/Evaluate, followed by Input Parameters:
+Examples, Store, Load and Reset. Store downloads
 `parameter.json` containing non-default settings and embedded input text in the
 [native JSON format](../../doc/json-parameters.md). `_comment` records compiled
 tool/library versions and executable reproduction commands. It is JSON metadata,
@@ -26,11 +27,48 @@ and CSV-only settings while retaining the same inputs and analysis settings.
 Reports run independently, matching the worker. The WASM integration test
 replays the JSON in fresh engines and compares every output byte.
 
+## Computed sessions
+
+Each successful computation, including a run with no interaction rows, creates a
+new session with a unique ID within the page and a default name such as
+`Prediction 1` or `Evaluation 2`. The session retains the computed inputs in JSON,
+CSV output, additional files and diagnostics. Editing parameters, Load, examples,
+Reset, invalid inputs, failed computations and cancellation preserve prior
+sessions. Failed or aborted computations do not create a session.
+
+The newest session opens automatically under **Results**. Use **Show**, select
+its header/name, or select its output area to restore that session’s
+computed inputs and fold the other sessions. The task, sequence text, signed
+indices, SHAPE/accessibility text, CSV selection/order and additional output
+selection are restored together. **Hide** folds the output. Each session
+keeps its own filter, sorting and visible columns; folded vaRRI frames are unloaded
+and restored when their session is reopened.
+
+Edit **Session name** directly in the header. Names are plain text; an empty name
+returns to the default. **Download ZIP** remains available while folded and always
+archives that session’s original JSON and complete outputs, regardless of the
+current form or table filters. The session name supplies a safe download filename;
+ZIP members remain `parameter.json` and output files. The garbage-bin button
+removes only its session and releases its output; it does not change form inputs.
+
+Sessions exist only in the current page’s browser memory. They are not written to
+browser storage or uploaded. Reloading or closing the page discards them; use ZIP
+exports to retain results. Memory use grows with the runs kept open, so delete
+sessions that are no longer needed. Store continues to save the editable form.
+If compiled metadata cannot be fetched, a completed run is still retained and its
+JSON explicitly records that version information is unavailable.
+
 ## Reset and examples
 
-Reset clears sequences, interactions, uploaded files and results and restores
-prediction defaults. Empty FASTA fields show the short demonstration sequences
-as placeholders only. The Example disclosure menu sits to the left of Reset.
+Reset clears the editable sequences, interactions and uploaded files and restores
+prediction defaults (10 interactions per pair, expanded energy/seed columns).
+Evaluation defaults select all energy columns. These GUI defaults are written
+explicitly in JSON when they differ from the CLI; importing existing JSON still
+uses native defaults for omitted parameters. Inactive controls keep their edits
+in the form for switching modes, but their values are omitted from executed and
+stored settings. Previously computed sessions remain available. Empty FASTA
+fields show the short demonstration sequences as placeholders only. The Examples
+disclosure menu follows the Input Parameters label.
 
 Maintainers edit `gui/examples/index.json` (an array of local JSON filenames).
 Each listed file contains `name`, `description`, and `configuration`, where the
@@ -42,9 +80,13 @@ examples show an error in the menu; they never overwrite the form.
 ## Local FASTA files
 
 Each strand has an Upload button and accepts a text-file drop into its textarea.
-Files are read locally (at most 60 KB), validated with the same FASTA checks as
+Files are read locally (at most 1 MB), validated with the same FASTA checks as
 manual entry, and replace only that strand after successful validation. Prediction
-accepts multi-FASTA; evaluation requires one record. Invalid files leave the
+accepts up to 50 FASTA records per strand, each 2–5,000 nt; evaluation requires
+one record. The longest selected query × longest selected target must be at most
+1,000,000. Names end at the first whitespace; ignored header descriptions produce
+a warning below the corresponding field. Both FASTA fields scroll horizontally
+and vertically. Invalid files leave the
 previous input intact. A later edit, reset, import or computation discards any
 pending file read, so delayed reads cannot replace newer input.
 
@@ -66,3 +108,42 @@ pending file read, so delayed reads cannot replace newer input.
 - Embedded ViennaRNA energy-parameter text matched native file input. Existing
   dependency-download and ViennaRNA-release tests, shell syntax, workflow YAML,
   source-distribution contents, vendor checksums and `git diff --check` passed.
+
+## Session follow-up validation (2026-10-23)
+
+- `npm test --prefix gui`: all 18 Node tests passed.
+- `scripts/wasm/test-browser.py` passed in Chromium and Firefox with real WASM
+  at nested URLs. Session checks cover input/JSON restoration, task changes,
+  signed indices, SHAPE text, CSV order, reports, independent filters/columns,
+  edits/reset/imports, failed/aborted runs, keyboard selection, safe renaming,
+  folded-session archives, deletion, stale file reads and page reload.
+- Desktop screenshots were inspected; session controls remain accessible at a
+  390 px viewport without document overflow.
+- `node scripts/wasm/smoke-protocol.mjs` passed native-reference checks and
+  byte-identical archive replay. Browser checks reused the parent branch’s
+  compiled Emscripten 4.0.23 / ViennaRNA 2.7.2 assets with their original metadata;
+  this follow-up changes no C++ code or worker protocol.
+- A fresh GCC 14 release build passed `make tests -j2`: all nine native API/CLI
+  groups passed. Tracked fixtures were linked into the out-of-tree build’s
+  `tests/data` directory for the existing CLI harness.
+- Static staging (including the new session module and engine digest checks),
+  shell syntax and `git diff --check` passed.
+
+## PR #299 review validation (2026-10-24)
+
+- `npm test --prefix gui`: all 20 tests passed, including FASTA limits, header
+  warnings, complete column grouping and explicit browser-default JSON replay.
+- `scripts/wasm/test-browser.py` passed with real WASM in Chromium and Firefox.
+  It covers the reordered panels, aligned controls, shared accessibility flags,
+  seed/accessibility/helix modes, inactive values, explicit-seed Load/Store,
+  group selection, FASTA scrolling/warnings and result header/name selection,
+  plus the existing prediction/evaluation, ZIP, abort and session checks.
+- Desktop and 390 px mobile screenshots were inspected. All seven panels have
+  no document overflow; table contents retain their own horizontal scrolling.
+- `node scripts/wasm/smoke-protocol.mjs <staged-engine>` passed native-reference
+  checks and byte-identical JSON archive replay. The unchanged engine uses the
+  parent branch's Emscripten 4.0.23 / ViennaRNA 2.7.2 assets and original metadata.
+- A fresh GCC 14 release build passed `make tests -j2`: all nine API/CLI groups.
+  Existing CLI fixtures were linked into the out-of-tree `tests/data` directory.
+- Static staging with engine digest validation, JavaScript/Python syntax checks,
+  shell syntax and `git diff --check` passed.

gui/docs/step3-frontend-ui.md

@@ -21,25 +21,26 @@ the group website. Footer and table notes use the available width, wrapping only
 when the viewport requires it.
 
 ```text
-Group logo       IntaRNA / RNA–RNA interaction analysis        University logo
+Group logo       IntaRNA package / RNA–RNA interaction analysis        University logo
 RNA–RNA interactions                             Computations stay on your device
-Task: Prediction / Evaluation                                Load example
-[ Sequence input | Query | Target | Seed | Helix | SHAPE | Energy | Output | General ]
+Task: Prediction / Evaluation
+[ Sequences | Seed | Output | Model & Energy | Query | Target | CLI-only ]
 [ Active panel: e.g. query + target, start indices, evaluation structures      ]
-Reset parameters                                                Compute
-Interactions / search / show-hide columns / CSV + ZIP download
+Compute     Input Parameters: Examples / Store / Load / Reset
+Results / Show-Hide / name / ZIP download / delete
+Interactions / search / show-hide columns / CSV download
 Selected interaction / embedded vaRRI
 Additional output files / diagnostics / compiled versions
 ```
 
-Every CLI group has a tab with one visible panel. Sequence input is selected
+Every CLI group has a tab with one visible panel. Sequences is selected
 initially. Tabs expose `aria-selected`, `aria-controls` and roving focus; Left/Right
 move between available tabs, and Home/End select the first/last. Tab names wrap
 on narrow screens. Prediction-only tabs and controls are disabled/hidden during
-evaluation; if the selected tab becomes unavailable, Sequence input is selected.
-Changing task resets parameters to their CLI defaults while preserving sequence text; Load example also restores the
-complementary pair. The native default output count is one; browser users can
-request 0–100 interactions per pair. Zero enables accessibility-only output.
+evaluation; if the selected tab becomes unavailable, Sequences is selected.
+Changing task resets parameters to their task defaults while preserving sequence text; Load example also restores the
+complementary pair. The browser defaults to 10 interactions per pair (the native default remains
+one), and accepts 0–100. Zero enables accessibility-only output.
 
 ## Inputs, defaults and validation
 
@@ -53,15 +54,29 @@ CSV uses a fixed semicolon delimiter.
 Process settings (threads, file/log paths, parameter files, help/version commands)
 are replaced by the worker lifecycle, local data inputs, downloads and displayed
 compiled versions. The base-pair maximization energy model is intentionally
-excluded. The helix-block interaction model B is supported.
+excluded. The helix-block interaction model B is supported, with helix controls at the end
+of Model & Energy, enabled only for model B. Seed constraint offers Disable,
+Explicit seed and Identify seeds, enabling only the relevant controls. Query and
+Target include their own accessibility/SHAPE columns and synchronized copies of
+the two shared accessibility flags. Ignore disables accessibility controls;
+import modes disable window/span and enable the data input. Disabled settings
+remain editable when re-enabled but do not enter the request. Explicit row grids
+align inputs even when their labels wrap.
 
-Prediction accepts plain sequence or up to 20 FASTA records per strand.
+CSV columns are grouped as Energy, Output, Sequence, Interaction and Partition
+Function, with per-group selection actions and name-only labels with description
+tooltips. Prediction adds ED1, ED2, seedE, seedStart1 and seedStart2 to the native
+CSV columns; evaluation selects all energy outputs. JSON records these browser
+defaults explicitly so native replay retains the same output.
+
+Prediction accepts plain sequence or up to 50 FASTA records per strand.
 Evaluation requires one query and target and 1–100 predefined structures.
 Wrapped/lowercase sequence text and DNA T normalize to uppercase RNA U. A/C/G/U/N
-are accepted. Each record is 2–2,000 nt; seed predictions additionally require
+are accepted. Each record is 2–5,000 nt; seed predictions additionally require
 at least `seedBP` nucleotides. FASTA IDs end at the first whitespace, matching the CLI. Duplicate IDs (including collisions after
-native CSV semicolon replacement) are rejected. The total selected all-vs-all
-query length × target length, or sum of pairwise products, is capped at 1,000,000.
+native CSV semicolon replacement) are rejected. Ignored header descriptions are reported below the input. The longest selected
+query length × longest selected target length is capped at 1,000,000, including
+pairwise mode; the cap is not summed over records.
 The per-strand start index applies to all records and defaults to one. Negative
 indices skip zero, matching `RnaSequence`; output coordinates remain inclusive.
 
@@ -78,7 +93,7 @@ accessibility text can be pasted or loaded from a local file (up to 1 MB). Data
 are written to fixed MEMFS paths, never used as user-provided CLI arguments.
 Native diagnostics remain available for model-specific failures.
 
-Inputs/results clear on edits; the form disables during execution. Abort and
-Escape terminate the worker during startup or C++ execution. Status, errors,
+Edits update the form while completed sessions retain their inputs and results;
+the form disables during execution. Abort and Escape terminate the worker during startup or C++ execution. Status, errors,
 empty output and diagnostic text have dedicated regions. Keyboard focus,
 reduced motion and mobile overflow are covered by real browser tests.