MAGE-TAB Object Model
The MAGE-TAB specification can be found at http://www.mged.org/mage-tab. A sample Molgenis implementation of the object model is at http://magetab-om.sourceforge.net. MAGE-TAB descriptions are actually bundles of three tab-delimited files (IDF, ADF and SDRF) with the (normally transcriptomics) data. MAGE-TAB does a good job at storing metadata for transcriptomics assays, that is also why it used as a standard for ArrayExpress submissions. General study information goes in the IDF (Investigation Description) file. Information about the array design goes in the ADF file. Finally, the SDRF file describes the relation between the samples as they result from study design and the actual data files.
Even though MAGE-TAB is a tabular format, it seems possible to describe the precise relations of the entities, as is done in the UML format at ftp://smd-ftp.stanford.edu/smd/transfers/magetab/. Below is a rendering of that: