qPCR Analysis App

Streamlit dashboard for quickly reviewing qPCR plates end-to-end – from pasted wells tables through replicate QC, standard curve fitting, absolute quantity estimation, ΔΔCt normalisation, and Excel export. The app is designed to work directly from the tables produced by most qPCR instruments, so you can validate a run without pre-processing in spreadsheets.

License: All Rights Reserved.

Status: this Streamlit app is the canonical version. A prior React/FastAPI experiment that lacked chart export has been removed to keep the repo lean.

Highlights

• Paste CSV/TSV/Excel-style well tables or vendor exports; delimiter detection and case-insensitive column matching keep the input flexible.
• Automatically reshapes vendor-specific columns, derives replicate numbers, and flags negative or control wells.
• Per-gene views surface replicate statistics (mean Cq, SD, ΔCq), standard curve plots with slope/intercept/R²/efficiency, and concentration back-calculations.
• Calibrator-aware normalisation plus relative expression via 2^-ΔΔCt with selectable reference genes and calibrator pools.
• One-click Excel export (qpcr_results.xlsx) that bundles cleaned wells, replicate summaries, standard curve points & plots, quantities, and relative-expression tables.
• Includes mock_wells.csv so you can try the workflow without real instrument output.

Screenshot

Example run using the bundled sample-data/qpcr_example.csv (select Example in the sidebar) and default options (refreshed Dec 28, 2025; lab notebook neo-brutalist styling applied):

Additional views from the full workflow:

Tests & screenshots

# Unit tests (core calculations)
./.venv/bin/python -m pytest

# E2E + screenshot refresh
npm install
npx playwright install --with-deps chromium
npm run test:e2e

Quick Start

1. Prerequisites: Python 3.9+ and pip. Creating a virtual environment is recommended.
2. Install dependencies

   python3 -m venv .venv
   source .venv/bin/activate  # Windows: .venv\Scripts\activate
   pip install -r requirements.txt

3. Run the dashboard

   streamlit run app.py

4. Streamlit prints a local URL (typically http://localhost:8501). Open it in your browser.

Desktop Installer (Windows)

From the repo root (next to this README):

npm install
npm run build:electron

The installer is generated in desktop/dist/ as an .exe (NSIS). On first run, the app asks for storage folders and creates them for you.

Notes:

• The packaged app expects Python 3.10+ available on PATH to run the Streamlit server. You can set APP_PYTHON_PATH to a specific Python executable if needed.
• The installer is unsigned unless code-signing credentials are configured.

Using the Dashboard

1. Input: Choose Example, Upload file, or Paste table in the sidebar. Example loads sample-data/qpcr_example.csv.
2. Column mapping: Confirm which columns correspond to Label/Cq/Replicate (helps when vendor exports don’t use standard headers).
3. Review & clean wells: Toggle keep to drop bad replicates; outliers by ΔCq are flagged.
4. Quantification mode: Pick Absolute (std curve) for standard-curve based quantities, or ΔΔCt (relative) for fold-change.
5. Export: Download an Excel workbook that bundles cleaned wells, summaries, curve points/plots (if applicable), and normalized outputs.

Use the download button at the bottom of the page to collect all derived tables (plus rendered standard-curve plots) into a single Excel workbook for record keeping.

Data format & quick AI helper

• Best-fit columns (CSV/TSV/Excel): Plate, Well, Gene, Type, Label, Replicate, Cq. Optional extras are ignored safely. See sample-data/qpcr_example.csv.
• To reshape quickly, use: ChatGPT, Gemini, Grok.
• Prompt: "Convert my table to CSV with headers: Plate, Well, Gene, Type, Label, Replicate, Cq. Keep my data, no invented rows, output UTF-8 CSV text only."
• Save as qpcr_example.csv (any name) and upload/paste. Visual helper: screenshots/data-format-helper.svg.

Tips

• Paste data exactly as exported; the app trims whitespace, harmonises case, and interprets common "NA"/"Undetermined" tokens automatically.
• Keep at least two standard levels with known concentrations per gene to enable curve fitting.
• When experimenting, use the Example input source to load sample-data/qpcr_example.csv (includes standards and samples).
• If you only want replicate QC (no standards), mock_wells.csv is a minimal wells table you can paste/upload.

How calculations are done

1. Clean & outliers – ΔCq per gene/label is median‐based; wells with ΔCq above the sidebar threshold are flagged. If a well is unchecked, its Cq becomes NaN.
2. Replicate collapse – Wells sharing Plate/Well/Gene/Type/Label are averaged (mean Cq) before any downstream math. Extra columns keep the first value.
3. Standards – For each gene (or Gene×Plate if selected):
   • log10(conc) from the standards map vs mean Cq → linear fit (slope, intercept).
   • R² and efficiency = (10^(−1/slope) − 1)·100.
4. Quantities (absolute) – For each kept sample well:
   • pred_log10Q = (Cq − intercept) / slope; Quantity = 10^pred_log10Q.
5. Normalisation to ref gene – Per label, RefQty is the mean Quantity of the chosen ref gene; Norm_Qty = Quantity / RefQty.
6. Exports – PerWell_Normalized carries one row per Plate/Well/Gene/Label with slope/intercept/pred_log10Q/Quantity/RefQty/Norm_Qty plus your metadata.

ΔΔCt option?

Yes — choose Quantification mode → ΔΔCt (relative) in the sidebar, then select calibrator labels. The app shows FoldChange (= 2^-ΔΔCt) and provides a CSV download.

Development Notes

• The Streamlit UI lives in app.py, while calculation helpers are in qpcr_core.py.
• Dependencies are listed in requirements.txt; dev-only tooling (pytest) is in requirements-dev.txt.
• Run unit tests with ./.venv/bin/python -m pytest; Playwright E2E refreshes the screenshots in screenshots/.