orange-assay 0.1.1

Orange Data Mining widgets for plate-image assays: microplate & coffee-ring readers and 3PL/4PL dose-response curve fitting (EC50, LOD).

orange-assay — plate-image & dose-response widgets for Orange Data Mining

English ｜ 中文

Three Orange Data Mining widgets that turn assay plate photos into tables and fit dose-response curves — without writing code. The image core reuses the coffee-ring-analyzer algorithm, so results match its Streamlit / mobile / CLI versions. Sibling package to orange-spectra.

Widget	What it does
Microplate Reader	Read a microtitre-plate photo (6–384 wells) into a per-well table: mean intensity, or thresholded count / intensity, with well labels (A1…H12).
Coffee-Ring Reader	Quantify a spotted-assay (coffee-ring) plate: per-cell area/intensity over a grid, blank-column (col 0) normalization and replicate averaging → a concentration / ratio table.
Dose-Response Fit	Fit a 3- or 4-parameter logistic curve to any concentration + signal table and report EC50, LOD and R², with the fitted curve overlaid on the points.

The outputs are ordinary Orange Tables, so they plug straight into Data Table, scatter plots, PCA, or the classification widgets — e.g. the Dose-Response Fit widget accepts either reader's output, or any colorimetric assay you already have in a spreadsheet.

Pipeline

Microplate Reader ─wells──┐
                          ├──► Dose-Response Fit ──► EC50 / LOD / curve
Coffee-Ring Reader ─cells─┘        (also: PCA · PLS-DA · Data Table)

Install

Works on Windows, macOS (Apple Silicon & Intel), and Linux — pure Python (numpy / scipy / matplotlib / Pillow / Orange3).

A. Desktop App (from orangedatamining.com): Options ▸ Add-ons… ▸ Add more…, type orange-assay, tick it, OK, restart.

B. pip Orange:

pip install orange-assay
python -m Orange.canvas

After restarting Orange, an Assay category with the three widgets appears in the toolbox. Each widget has an ℹ How to use box, and F1 opens its online help page. Full tutorial: https://tai-shengyeh.github.io/spectraview/assay.html (English).

Traditional-Chinese Windows note: if the Add-ons dialog crashes with UnicodeDecodeError: 'cp950' codec…, install from a command prompt instead: "…\Programs\Orange\python.exe" -m pip install orange-assay, then restart Orange (or set the PYTHONUTF8=1 environment variable).

Quick start

1. Microplate Reader → choose a plate photo → pick the format (e.g. 96) → Wells output → Data Table.
2. Or Coffee-Ring Reader → choose a spotted-plate photo → set rows × columns (column 0 = blank) → enter the concentrations → Cells output.
3. Feed either output to Dose-Response Fit → pick the concentration and signal columns → read EC50 / LOD / R² and the fitted curve.

Tests

python tests/test_core.py                                   # pure logic
QT_QPA_PLATFORM=offscreen python tests/test_widgets.py      # widget smoke tests

License

MIT.

---

中文說明

orange-assay — Orange Data Mining 盤面影像與劑量反應 widgets

English ｜ 中文

三個 Orange Data Mining widgets，把檢測盤照片 變成資料表並擬合劑量反應曲線——全程拖拉、免寫程式。影像核心重用 coffee-ring-analyzer 的演算法，結果與其 Streamlit／手機／CLI 版本一致。與 orange-spectra 為姊妹套件。

Widget	功能
Microplate Reader	讀微孔盤照片（6–384 孔）成每孔一列的表：平均亮度，或閾值面積／強度，附孔位標籤（A1…H12）。
Coffee-Ring Reader	量化點樣（咖啡環）盤：格點逐格取面積／強度、除以同列 blank（第 0 欄）、平均重複列 → 濃度／比值 表。
Dose-Response Fit	對任何 濃度＋訊號 表擬合 3 或 4 參數邏輯曲線，回報 EC50、LOD、R²，並把擬合曲線疊在資料點上。

輸出都是一般的 Orange Table，可直接接 Data Table、散佈圖、PCA 或分類 widget； Dose-Response Fit 接受兩個 reader 的輸出，也吃你已在試算表裡的比色法數據。

工作流程

Microplate Reader ─wells──┐
                          ├──► Dose-Response Fit ──► EC50 / LOD / 曲線
Coffee-Ring Reader ─cells─┘        （也可接 PCA · PLS-DA · Data Table）

安裝

可在 Windows、macOS（Apple 晶片與 Intel）、Linux 執行——純 Python （numpy / scipy / matplotlib / Pillow / Orange3）。

A. 桌面版 App：Options ▸ Add-ons… ▸ Add more… 輸入 orange-assay → 打勾 → OK → 重啟。

B. pip 版 Orange：

pip install orange-assay
python -m Orange.canvas

重啟後工具箱會出現 Assay 分類（3 個 widgets）。每個 widget 都有 「ℹ 說明」盒子，按 F1 會開線上說明頁。教學網頁：https://tai-shengyeh.github.io/spectraview/assay.html（English）。

繁體中文 Windows 注意： 若 Add-ons 對話框出現 UnicodeDecodeError: 'cp950' codec… 崩潰，改用命令列安裝： "…\Programs\Orange\python.exe" -m pip install orange-assay，再重開 Orange （或設環境變數 PYTHONUTF8=1）。

快速上手

1. Microplate Reader → 選盤面照片 → 選規格（如 96）→ Wells 輸出 → Data Table。
2. 或 Coffee-Ring Reader → 選點樣盤照片 → 設列×欄（第 0 欄＝blank）→ 填濃度 → Cells 輸出。
3. 任一輸出接 Dose-Response Fit → 選濃度與訊號欄 → 讀 EC50／LOD／R² 與曲線。

測試

python tests/test_core.py                                   # 純邏輯
QT_QPA_PLATFORM=offscreen python tests/test_widgets.py      # widget 煙霧測試

授權

MIT。