Free, open-source dose-response curve fitting and IC50 calculator — replaces the GraphPad Prism workflow.
DoseCurve fits a 4-Parameter Logistic (4PL) model to your concentration-response data and reports the IC50 with 95% confidence intervals — the same analysis pharmacologists pay $142–$1,040/yr to run in GraphPad Prism. Paste your data, click Fit, get publication-quality results in seconds.
- Paste-and-go data entry — paste concentration-response pairs or upload CSV, with replicate support
- 4PL curve fitting — Levenberg-Marquardt nonlinear least-squares regression
- IC50 + confidence intervals — IC50, Hill slope, top/bottom plateaus with 95% CI via Jacobian-based covariance
- Publication-quality chart — log₁₀ concentration x-axis, data points with SEM error bars, fitted curve overlay, IC50 annotation
- Goodness-of-fit metrics — R², sum of squared residuals, residual plot
- Export — download chart as PNG/SVG, results as CSV
- Light & dark theme
# Prerequisites: Node.js ≥ 18, pnpm
git clone https://github.com/your-org/dosecurve.git
cd dosecurve
pnpm install
pnpm build # build engine + web
pnpm dev # start dev server
pnpm test # run all testsOpen http://localhost:5173, paste your concentration-response data, and hit Fit Curve.
y = Bottom + (Top − Bottom) / (1 + (x / IC50)^n)
| Parameter | Meaning |
|---|---|
| Top | Upper asymptote (max response) |
| Bottom | Lower asymptote (min response) |
| IC50 | Concentration producing 50% response |
| n (Hill slope) | Steepness of the curve |
Iteratively solves for parameter updates:
Δp = (JᵀJ + λI)⁻¹ Jᵀr
where J is the Jacobian matrix, r is the residual vector, and λ is the damping factor that interpolates between gradient descent and Gauss-Newton.
R² = 1 − SS_res / SS_tot
| Layer | Technology |
|---|---|
| Curve fitting engine | TypeScript (pure math, zero dependencies) |
| Web UI | React 18 + Vite |
| Charting | Recharts |
| Monorepo | pnpm workspaces |
| Testing | Vitest |
dosecurve/
├── packages/
│ ├── engine/ # 4PL fitting library
│ │ └── src/
│ │ ├── parser.ts # data parsing & replicate summarization
│ │ ├── fit.ts # 4PL model + Levenberg-Marquardt solver
│ │ ├── confidence.ts # 95% CI via Jacobian covariance
│ │ ├── goodness.ts # R², residuals
│ │ ├── export.ts # CSV + curve-point generation
│ │ └── __tests__/
│ └── web/ # React front-end
│ └── src/
│ ├── components/
│ │ ├── DataEntry.tsx
│ │ ├── DoseResponseChart.tsx
│ │ ├── ResultsPanel.tsx
│ │ └── Toolbar.tsx
│ └── App.tsx
├── PLAN.md
├── package.json
└── pnpm-workspace.yaml
MIT