SciDAVis for Life Sciences & Pharma: Dose–Response, Kinetics, and Descriptive Stats

Plate readers, enzyme assays, cell counts, growth curves — life-science data is mostly small tables and big questions: what is the IC50, does this enzyme follow Michaelis–Menten kinetics, are these groups really different. SciDAVis (free, open-source; latest release v2.9.2, precompiled binaries on SourceForge are v2.8) covers the core of wet-lab statistics: sigmoidal dose–response fits, custom kinetic models, histograms and box plots, and one-click descriptive statistics. This guide shows each workflow and states plainly where the tool stops.

Version note: written for v2.9.x (SourceForge binaries are 2.8). The statistics and fitting commands described here are unchanged between the two.

Step 1 — Dose–response curves and IC50/EC50

Enter concentration and response, plot the curve, then Analysis → Quick Fit → Fit Boltzmann (Sigmoidal). The fitted sigmoid’s midpoint parameter is the EC50 (or IC50 for inhibition) concentration — read it from the fit log together with its standard error. Practical notes: plot against log(concentration) when the doses span orders of magnitude; the fit handles the scale either way. Use Graph → Add Error Bars… to show the spread of replicates — a dose–response without replicate error bars is a claim without evidence. The slope parameter of the sigmoid (Hill slope) is worth reporting when comparing compounds — a slope far from 1 hints at assay artifacts or cooperative binding.

Then check the residuals by eye: zoom the graph and confirm they scatter randomly around zero. A systematic wave means the response is not sigmoidal (biphasic response, partial agonism) and no single IC50 is meaningful. The fitting mechanics, including how to read the results log and judge a fit, are in the curve fitting tutorial.

Screening several compounds? Fit each dose–response in its own table, then collect the midpoints and standard errors into a summary table — one row per compound — for a clean potency ranking. Keep the raw curves in the project: a ranked table without the underlying fits is not auditable, and reviewers will ask.

Step 2 — Enzyme kinetics without a built-in model

Honest statement first: SciDAVis has no built-in Michaelis–Menten fit. The route is Analysis → Fit Wizard (Ctrl+Y) with a custom formula — for example v = Vmax*x/(Km+x) — where you name the parameters, supply starting guesses, and optionally restrict the x-range. Good starting guesses come free from the data: Vmax near the highest observed rate, Km near the substrate concentration giving half that rate. The fit returns Vmax and Km with standard errors, and it is the statistically honest route: fit the raw data, not the linearized Lineweaver–Burk transform (linearization distorts the error structure, so its confidence intervals mislead). If a reviewer asks for the double-reciprocal plot anyway, make it as a display figure computed from the fitted parameters — not as the fitting route.

Residual checks apply here too: curvature in the residuals of a Michaelis–Menten fit is how you discover substrate inhibition or cooperativity — the model failing is data, not an error.

Step 3 — See the distribution: histograms and box plots

Before any test, look at the data. Select your data columns in the table, then Plot → Statistical graphs → Histogram builds a frequency histogram — the default binning uses 10 steps between the min and max of the Y-values and can be adjusted in the Custom curves dialog — ideal for cell-size distributions or replicate spread. Plot → Statistical graphs → Box Plot summarizes each column as a box (mean, min/max, 5th and 95th percentiles), which is the fastest honest comparison of treated versus control groups. If the box plots of two groups barely overlap, you have an effect worth testing properly; if they overlap completely, no test will save the story. Comparing several columns at once works too: Statistical graphs → Stacked Histogram draws one histogram per layer in vertically stacked panels, handy for scanning replicate batches for an outlier run.

Step 4 — Descriptive statistics in one click — and what is missing

Select the column(s) in the table and choose Analysis → Statistics on Columns. SciDAVis creates a new table, “Column Statistics of <your table>”, with one row per column: mean, standard deviation, variance, min, max, N, and more. Analysis → Statistics on Rows does the same across rows, useful when each row is one experimental replicate. Only numeric columns are summarized; empty or invalid cells are ignored rather than treated as zeros — the N column tells you how many valid values were actually used, so check it before quoting any mean. This is the entire built-in statistical capability of the program — and here is the honest boundary: SciDAVis 2.9.x has no t-test, no ANOVA, and no normality test. For those, export the table as CSV and run them in Python/SciPy (SciDAVis has a documented Python scripting API for the repetitive parts), R, or JASP — all free. Do not describe p-values from a tool that cannot compute them. The statistics tutorial documents exactly what the statistics commands produce and the workarounds for the missing tests.