Back to Biology calculators

Assay standard curves guide

Use this page when the real question is how to interpret a protein assay or ELISA curve: blank handling, fit choice, range checks, and what to write in the report after you run the calculator.

Other languages 日本語 | English | 简体中文 | Español | Português (Brasil) | Bahasa Indonesia | Français | हिन्दी | العربية
Protein standard curve ELISA standard curve Biology hub

When this guide is more useful than the calculator alone

If you already know which curve model to run and only need the fitted concentration, go straight to the calculator. Stay on this guide when the harder question is whether the blank should be subtracted, whether the fitted shape is defensible, or whether your unknown should be treated as an extrapolation instead of a reportable result.

Protein assay vs ELISA: where the workflow splits

Protein assays such as BCA and Bradford often behave like narrower-range standard curves, so the practical choice is usually between a linear and a quadratic fit plus consistent blank handling. ELISA workflows are more often sigmoidal, so the real decision shifts toward 4PL vs 5PL, weighting, and whether the upper and lower plateaus are behaving sensibly.

Blank subtraction and baseline decisions

Blank subtraction is not an automatic improvement. Use it when the blank captures the same background signal that standards and unknowns carry, and keep the treatment consistent across the whole plate or run. If subtracting the blank changes the story too much, that is often a cue to inspect the assay setup rather than a reason to force the cleaner-looking answer.

For protein assays, pay attention to whether the zero standard really behaves like the unknown matrix. For ELISA, decide whether the 0 concentration row belongs in the fit, should be subtracted then excluded, or should stay outside the fit on a log-x workflow.

Choosing the fit model

Model choice should follow assay behavior, not preference for a higher-looking score. Linear and quadratic models can be appropriate for protein assays across modest ranges. 4PL and 5PL are better fits when the assay response is clearly sigmoidal and the plate spans upper and lower plateaus.

  1. Start with the simplest model that matches the assay family.
  2. Check residual patterns, not just one summary metric.
  3. Only move to the more flexible model when it fixes a real mismatch rather than cosmetic noise.

Range checks, outliers, and extrapolation

Unknowns should ideally sit inside the bracketed standard range. If an unknown lands outside, the number becomes an extrapolation warning. That is often the right time to adjust dilution, rerun the sample, or tighten the standard range rather than accepting the first estimate as stable.

A single suspicious point also matters more than a smooth-looking headline metric. Outlier treatment should be explicit: say what was excluded, why, and whether the conclusion changes if that point stays in.

What to write in the report

A reproducible assay write-up needs more than the final concentration. Include the assay type, standard range, fit model, weighting if any, blank handling, excluded points, dilution factors, and whether any unknown was extrapolated. Those decisions explain why another person would trust your concentration estimate.

Recommended next pages

FAQ

When should I subtract the blank?

Subtract the blank when the assay design and plate setup support one shared baseline for standards and unknowns. Do not subtract it just to force a cleaner-looking curve; first confirm that the blank truly represents the same background signal your unknowns carry.

How do I choose between linear, quadratic, 4PL, and 5PL?

Use linear or quadratic fits when the assay behaves like a narrower-range standard curve, such as many BCA or Bradford workflows. Use 4PL or 5PL when the assay is sigmoidal across a wider range, as in many ELISA workflows. Pick the simplest model that still matches the residual pattern and the assay design.

What should I do when an unknown falls outside the standard range?

Treat it as an extrapolation warning, not a stable result. Dilute or concentrate the sample, remeasure within range, and report that the first estimate sat outside the bracketed standards if you need to keep the record complete.

What must I report so the assay analysis is reproducible?

Report the assay type, standard range, fit model, weighting if used, blank handling, exclusions or outlier treatment, unknown dilution factors, and whether any value was extrapolated. Those details matter more than the final concentration alone.