OpenDDA: LFQ/TMT quickstart, decoy detection, PSM FDR level, installer link

[t0mdavid-m]

Requested by Tom · project thread

Before: the quickstart described only label-free quantification. The "Download for Windows" button served the streamlit-template installer, because the Docker images pulled OpenMS-App.zip from streamlit-template releases. Fragment tolerance could only be entered to two decimals. Local installs showed the browser upload widget. Decoy generation was always on, even for FASTA files that already contain decoys. The PSM FDR could only be set through IDFilter's raw score fields.

After:

• Quickstart and app name present OpenDDA with LFQ and TMT workflows side by side.
• Installer: all four Dockerfiles download from OpenMS/quantms-web releases (the OpenDDA-* releases carry OpenMS-App.zip). repository-name and the Windows Readme link point at quantms-web.
• Decoys: "Generate Decoy Database" stays on by default, but turns off when the selected FASTA already has DECOY_, decoy_, rev_, REV_, XXX_ or reversed_ entries. The detected prefix becomes Comet's decoy_string. This happens once per database, so a user's own choice is kept. TMT Percolator and ProteinInference now receive the upstream decoy string as well; before, they hard-coded DECOY_ while the generated decoys used rev_.
• New PSM FDR level (%) setting on the Filtering tab (LFQ and TMT). It sets IDFilter's q-value cutoff, and at 100% the filter is skipped so PSMs pass through unfiltered. The default is 1%; the previous IDFilter default was 10%. The ProteomicsLFQ and protein-level FDRs keep their own settings.
• Ported from Full-precision float inputs; local installs pick files from disk streamlit-template#405: full-precision float inputs and local-only file picking.

The Abundance page's duplicated tab row, which made the PSM tab show the "Protein-Level" heading, is fixed in #36. That PR rewrites the page, so this PR leaves content/results_abundance.py alone to avoid a conflict. Both PRs share src/WorkflowTest.py, so whichever merges second may need a merge from main.

How: decoy detection is a cached scan of the FASTA headers (detect_decoy_prefix). apply_decoy_default() runs from configure(), and the FASTA selector is now reactive so it follows changes. filter_psms() wraps IDFilter for both modes.

Notes: repository-name also names the local workspace folder, which moves from ../workspaces-streamlit-template to ../workspaces-quantms-web. Workspaces from earlier local runs will not show up automatically.

Tests: python -m pytest test_gui.py tests/ passes (102). The upload and configure pages render under AppTest. Decoy auto-detection was checked with a FASTA containing DECOY_ entries: the checkbox turns off, decoy_string is set to DECOY_, and a later manual override is kept.

🤖 Generated with Claude Code

https://claude.ai/code/session_01C37vTrifAfq9uazSL7UUAR