simpleredcapbuilder

Background

REDCap is a fabulously useful tool for creating web databases, including a useful scheme for defining the database schema ("data dictionaries") in plain CSV files. However, difficulties arise when designing complex schema. It is cumbersome to code multiuple occurences of essentially the same field (e.g. sample_1 ... sample_2 ...). If changes need to be made, keeping wording and behaviour consistent across multiple instances is tedious. If a number of similar databases have to be deployed, the consistency problem is multipled.

simpleredcapbuilder allows you to write REDCap schema in a more compact form that allows for repetition of rows, sections and forms (instruments), with straightforward text substitution into the repeats. It allows make checks on the schema, looking for possibly invalid values.

Installation

simpleredcapbuilder can be downloaded from the repo and installed in the usual sort of ways:

• Download the repo, unpack it if necessary and use setup.py:

  % python setup.py install
  

• Use a package manager pointed at the git repository:

  > pip install git+https://github.com/agapow/simpleredcapbuilder.git
  

simpleredcapbuilder has been built under Python3 but should run under Python2.

Usage

simpleredcapbuilder installs a single script, that is called:

usage: expand-redcap-schema [-h] [-o OUTFILE] infile

positional arguments:
  infile                compact REDCap file to be processed

optional arguments:
  -h, --help            show this help message and exit
  -o OUTFILE, --outfile OUTFILE
                        output expanded redcap data dictionary

This accepts a "compact" REDCap data dictionary, which allows which actually follows the form of the standard data dictionary with two additional columns: tags and repeat. (Note: the tags column is not currently used.). The repeat column says whether the row, or section or form starting with that row, should be repeated. For example:

• 1-7
• item: 'A', 'B', 'C'
• section: 'foo', 'bar', 'baz'
• form: (1, 'one'), (2, 'two'), (3, 'three')

That is, the repeat element starts with a qualifier saying whether the repeat is for the row, section or form. If no qualifier is given, it is assumed to be for the item. This is followed by either a range of numbers, or a sequence of values. (The sequence is anything that when enclosed in square braces will look like a legal list to Python.)

When this data dictionary is expanded, the associated items will be repeated. See the example simple-but-useless for an illustration.

Obviously, simply repeating a row or section is useless: variable identifiers will be repeated, fields will have the same name, etc. simpleredcapbuildergets around this problem by embedding and interpreting a template language (jinja2) in the data dictionary. For example, an item may have the repeat value of item: 1-3 and the variable name of sample_{{ i_iter }} and the field label of Sample number {[ i_iter ]}. These will result in three rows, named and labelled:

sample_1 ... Sample number 1
sample_2 ... Sample number 2
sample_3 ... Sample number 3

The sequence of values used in any repeat loop are i_iter, s_iter and f_iter for item, section and form repeats respectively. See the example simple-and-useful for an illustration.

More complex transformations are possible and you are referred to the Jinja documentation.

References

• simpleredcapbuilder code repository
• Jinja2 templating system