Sparrow helps to make your lab data findable, accessible, interoperable, and reusable. To do this, your data needs to be in a form that can be passed to the Sparrow database. This means you will likely need to minimize variability in your files that humans may have added. Some useful information on the source of samples and papers may not be easily accessible in the archive you have, but Sparrow will help with that later.
You should plan on iterative development of an import pipeline, especially if you have a large dataset that may include variation from human input. For instance, if you have 3 common methods, start an import pipeline that extracts data for one of them and then build on it. It is difficult to build an automation that works for an entire archive the first time, but be persistent and work toward completeness later.
- Clean your analytical data files to make uniform columns and cells.
- Iterate to ingest as much of your archive as possible.
- Link metadata to analytical data in Sparrow.
- Share data to the outside world!
Uniform column headers and data types in cells are two of the most important steps in organizing and cleaning data from Excel files. This may be as simple to complete as writing a single regular expression to find all columns with a similar form (see this example website). Simply, the minimum matching structured characters may be what is necessary to always find the needed columns. If you want more control, make your regular expression strings longer, or use a look-up dictionary with all possible columns listed (see WiscSIMS importer example.
For data in cells, it is important to ensure that there are no
where not allowed in the Sparrow import pipeline. This can be handled in the
steps to transform tables into JSON. See the schema for more information on what can not be null.
Start simple and build complexity is the fundamental rule for building importers. Unless your initial files are well structured and have minimal variability from user input, it is unreasonable to expect 100% success on the first attempt at importing data. Identify the minimal data necessary to populate your instance of Sparrow with findable, accessible, interoperable, and reusable data. For instance, if standards run with samples and the values of both are necessary to check for the accuracy and precision of data in your lab, make sure to import both.
Writing data importers for Sparrow is an iterative and long yet rewarding process that makes your lab's data findable, accessible, and interoperable.
Once data is cleaned and structured through this iterative process, Sparrow provides an easy
function for importing data,
db.load_data. Examples, below, show this function in more detail.
Two examples for importers can be found: