Measurements and data¶

The Experimental Measurements and Data tab holds the numbers a model will be scored against. It has three linked columns: measurements on the left, the data sets belonging to the selected measurement in the middle, and the data points of the selected data set on the right.

Measurements¶

A measurement is one observable quantity — "cytosolic ATP", "labelled fraction of plasma glucose", "GFP intensity in the nucleus".

Press Add to create one. The dialog offers the measurements that already exist in the database as well as a new one, so a quantity measured across several experiments can stay a single measurement.

The important field is the Variable: the model quantity this measurement corresponds to. Setting it is what lets ProcessDB plot your data on the same axes as the simulation and compute residuals. Measurements without a variable are still stored; they simply take no part in fitting.

Data sets¶

A data set is one time course: one replicate, one animal, one condition. Measurements usually have several.

Each data set has a weight, which scales its contribution to the objective function relative to the others. Use it when some replicates are more trustworthy than others; leave it alone if they are equivalent.

Data points¶

Each data point has a Time, a Value, and two weights — a Time Weight and a Value Weight — which let individual points count for more or less in a fit. You can type points in directly, but for real data you will import them.

Weights are usually set in bulk rather than one point at a time. See Fitting to data for the weighting formulas available from the Parameter Explorer.

Importing from Excel¶

Select exactly one data set and press Import.

1. Browse to an .xlsx or .xls file.
2. Enter the sheet name, or press Select… to pick the sheet and drag out the region holding your data — including the + row at the bottom if you want to read to the end of the column.
3. Check the Data Preview. ProcessDB expects times in one column and values in another; if it has read your sheet sideways, press Transpose.
4. Press OK.

The points land in the selected data set. As with everything else, they are local until you commit.

What happens to the data next¶

Data reaches a simulation through the Model of Experiment:

• In the Parameter Explorer, the data belonging to a measurement is drawn on any graph showing the corresponding model variable, so you see model and measurement together.
• In Edit Associations and Weights, you can review and change which model variable each measurement maps to, and set weights in bulk.
• The optimizer minimises the weighted sum of squared residuals between the two.

If your data does not appear on a graph, the usual cause is a measurement with no Variable set, or a variable name that no longer exists in the model.