Skip to content

Metadata File Format: Clinical Variables

Mark Keller edited this page Feb 2, 2019 · 5 revisions

The clinical variables metadata file specifies the variables (and their values) available in the clinical data files. (It is fine if the clinical data files contain more columns than those specified in this file, those columns will just be ignored.)

It must be located at obj/meta-clinical.tsv

The clinical metadata file contains the following columns:

  • Clinical Column: The column name, which will also be the name of the clinical variable in ExploSig.
  • Scale Type: Either continuous or categorical.
  • Extent: Whether or not to infer the "extent" of the variable values (the possible values the variable can take). If the extent should be inferred, the column value should be infer and the Value column for this row should remain empty. If not, the Extent column value should remain blank and the Value column should be used.
  • Value: If not inferring the extent of a variable, this column specifies which values the variable can take. If the variable is continuous, there should be two rows, where in the first row this column contains the minimum value and in the second row this column contains the maximum value. If the variable is categorical, there should be as many rows as there are possible values.

Note that clinical variable values are assumed to be human-readable (at the expense of sometimes being verbose) and that there is no conversion mechanism between the values in the clinical data files and the values shown to the user in ExploSig. For example, rather than storing a "binary" variable like "Vital Status" as 0 and 1 for dead and alive, just store the strings "Dead" and "Alive".

An example of this file looks as follows:

Clinical Column Scale Type Extent Value
Diagnosis Age continuous 0
Diagnosis Age continuous 100
Sex categorical Male
Sex categorical Female
Overall Survival continuous infer