<aside> 📊

Reading in data, normalizing it, and fitting curves. Think carefully about your data and pick a combination and order of these normalization methods that suits the finding you're trying to show — they are not meant to all be applied every time.

</aside>

Conventions

Column and dataframe names should be lowercase with no spaces.
Throughout this page, replace phenotype with the column for your assay:

Assay	Column
Motility	`optical_flow`
Viability	`AreaOccupied_AreaOccupied_GreenWorms`
Development	`AreaShape_Area`, `AreaShape_MajorAxisLength`, etc.
Fecundity	`pixel_count`

Reading in data

The raw dataframe is df.raw.

From an RDS file
From multiple CSV files

Concentration conversion and dose-response curves

If you are fitting a dose-response curve, convert concentration values to numbers first.

convert_conc() — string concentrations to molar
fit_dose_response() — baseline curve fit returning ED50
Curves for multiple groups at once

Outliers

Two approaches, with a real trade-off between them.

	Option 1 — separate dataframes	Option 2 — flag column
Produces	`df.trimmed` and `df.outliers`	`df` with an `outliers` column, TRUE/FALSE
Pro	Other normalization and plotting is easier	No data is lost; plots with and without outliers compare easily
Con	Lots of dataframes gets messy	Can skew other normalization schemes like Drift