Dose-response

Dose-response analysis fits a curve, usually a four-parameter logistic, to response against log concentration; the IC50 or EC50 is only meaningful with the curve, its confidence interval and the doses that define its top and bottom. A fit cannot invent plateaus the doses never reached. Space doses to cover both plateaus and the steep region, run replicates at each dose, and fix curve bounds only with a stated reason. Report the model, constraints, Hill slope and confidence interval, and compare curves with an appropriate test rather than by whether intervals overlap by eye.

A compound that never reaches half-maximal response has no IC50, however the software labels the output. Start by writing the question the Dose-response work must answer, the samples that can answer it, and the controls that would prove the answer wrong. Decide acceptance criteria at this stage, while changing them is still cheap. Pilot work belongs here: run the smallest version that exposes the main failure modes, look at the raw output before any summary, and adjust the design rather than the interpretation.

Keep a run record from the first pilot, with reagent lots, instrument settings, operator, date and every deviation, because the difference between a repeatable method and an anecdote is usually in those details. When the pilot passes its own criteria twice, freeze the protocol version you will use for the real samples. Quality control in Dose-response is a set of gates, not a final glance. Check controls first and stop if they fail; no sample result can rescue a failed control.

Then check the samples against criteria set in advance, such as replicate agreement, signal range, or completeness of the record, and document every exclusion with its reason. Where the method produces images, curves or plate maps, inspect the spatial and raw forms, because summaries hide gradients, saturation and single bad replicates. Keep failed runs in the record rather than deleting them. A failure log that shows what went wrong, how it was detected and what changed is strong evidence that the passing runs can be trusted.

Report Dose-response results so another scientist could repeat both the work and the calculation. State the method and version, materials with identifiers and lots, instrument and software settings, the exact analysis steps in order, and the statistical treatment with its assumptions. Show individual replicate values alongside summaries, give numbers with units, and distinguish exploratory findings from confirmatory ones. Deposit or archive the raw files, the analysis script or settings, and the final tables together under one identifier.

If a claim about BioStudioAI appears on a hub page, it must be checkable in the platform itself; where this hub does not yet have a dedicated guide, protocol or calculator for a sub-topic, the section hubs list what is available and this overview remains the starting point. Treat every new operator, instrument, reagent lot and site as a change that needs a short bridging check against the previous condition, and file that check with the run record so later readers can see continuity. Record who reviewed the analysis as well as who performed it, and keep that review dated and attached to the exact version of the results it covered.

Pages on Dose-response

Guides 9

Tools & calculators 3

Comparisons 1

Software reviews 1