Workflows
Dose-Response Analysis
Calculate IC50/EC50 with 4-parameter and 5-parameter logistic curve fitting
Dose-response analysis determines the potency of a compound by measuring how biological response changes with concentration. Gradiance supports both inhibition (IC₅₀) and stimulation (EC₅₀) assays.
When to Use Dose-Response Analysis
- Drug potency assays: Determine IC₅₀/EC₅₀ of compounds
- Cell viability tests: Cytotoxicity and proliferation assays
- Receptor binding: Agonist/antagonist screening
- Enzyme inhibition: Inhibitor characterization
Curve Fitting Models
4-Parameter Logistic (4PL)
The 4PL model is the standard for most biological assays. It fits a symmetric sigmoidal curve with four parameters:
- Top: Upper plateau (maximum response)
- Bottom: Lower plateau (minimum response)
- IC₅₀/EC₅₀: Concentration at 50% response
- Hill Slope: Steepness of the curve
When to use 4PL
Use 4PL when your curve is symmetric around the inflection point. This is appropriate for >90% of dose-response assays.
5-Parameter Logistic (5PL)
The 5PL model extends 4PL with an asymmetry parameter, allowing for non-symmetric curves. It includes all 4PL parameters plus:
- Asymmetry (S): Controls curve asymmetry around IC₅₀
When to use 5PL
Use 5PL when your curve is asymmetric (different slopes on each side of IC₅₀). Common in certain receptor assays and biphasic responses.
Configuration Options
Parameter Constraints
You can fix parameters to known values:
- Fix Top to 100%: When you know the maximum response (e.g., 100% viability with no drug)
- Fix Bottom to 0%: When you know the minimum response (e.g., complete inhibition)
Axis Transformations
- X-Axis (Concentration): Log10 scale (default) or Linear
- Y-Axis (Response): Linear (raw values) or Normalized (%)
Outlier Detection
- Auto-detect: AI identifies potential outliers using Z-score
- Manual exclusion: "Exclude point at 100 µM"
Quality Control Checks
Gradiance automatically runs QC checks on your fit:
| Check | Criteria | Severity |
|---|---|---|
| R² (goodness of fit) | < 0.95 | Warning |
| IC₅₀ extrapolation | IC₅₀ outside tested range | Warning |
| Low dynamic range | Top-Bottom < 50% of data range | Warning |
| Shallow slope | |Hill slope| < 0.5 | Warning |
Example Analysis
User Request
"Analyze this cell viability data and calculate IC₅₀ using a 4PL fit. The concentrations are in µM."
AI Response
✓ Detected concentration column: "Concentration_uM"
✓ Detected response column: "Viability_Percent"
✓ Fitted 4PL model
IC₅₀ = 2.34 µM (95% CI: 2.01 - 2.72)
R² = 0.987
Hill Slope = -1.12
Interactive Refinement
After the initial fit, you can refine your analysis:
- "Try a 5PL fit instead" - Switch to asymmetric model
- "Exclude the point at 100 µM" - Remove outlier
- "Fix the bottom to 0" - Constrain baseline
- "Show residual plot" - Check fit quality
Best Practices
- ✓ Use at least 6-8 concentration points spanning the IC₅₀
- ✓ Include technical replicates (3+ per concentration)
- ✓ Test concentrations covering 2-3 orders of magnitude
- ✓ Include vehicle control and maximum inhibition control
- ✓ Check R² > 0.95 for publication-quality fits