The information requirements of biological research change rapidly and are often unique to a particular experimental procedure. The LabKey Server experiment framework is designed to be flexible enough to meet these requirements. This flexibility, however, means that the purpose of an experiment description needs to be determined up-front, before creation of a xar.xml.
For example, the granularity of experimental procedure descriptions, how data sets are grouped into runs, and the types of annotations attached to the experiment description are all up to the author of the xar.xml. The appropriate answers to these design decisions depend on the uses intended for the experiment description.
Uses of the Experiment Framework
One reason to describe an experiment in xml is to enable the export and import of experimental results. If this is the author's sole purpose, the description can be minimal—a few broadly stated steps.
The experiment framework also serves as a place to record lab notes so that they are accessible through the same web site as the experimental results. It allows reviewers to drill in on the question, "How was this result achieved?" This use of the experiment framework is akin to publishing the pages from a lab notebook. When used for this purpose, the annotations can be blocks of descriptive text attached to the broadly stated steps.
A more ambitious use of experiment descriptions is to allow researchers to compare results and procedures across whatever dimensions they deem to be relevant. For example, the framework would enable the storage and comparison of annotations to answer questions such as:
- What are all the samples used in our lab that identified protein X with an expectation value of Y or less?
- How many samples from mice treated with substance S resulted in an identification of protein P?
- Does the concentration C of the reagent used in the depletion step affect the scores of peptides of type T?
In order to turn these questions into unambiguous and efficient queries to the database, the attributes in question need to be clearly specified and attached to the correct element of the experiment description.