Pygr is a Python graph database framework for bioinformatics; it recasts bioinformatics problems as mappings and graphs. It emphasizes Python's basic mapping model as a general interface for all of bioinformatics' diverse data types, while showing that this model can be highly scalable to huge comparative genomics datasets. Worldbase is Pygr's virtual namespace for scientific data sharing and publication. It enables users to save and access complex datasets (and their inter-relations) simply by its name within this virtual namespace. All aspects of finding the data, importing the right code to work with it, binding its schema attributes that connect it to other datasets etc. happen automatically. I'll illustrate how this greatly facilitates sharing of complex datasets in a variety of applications. Talk presented at the Data Analysis and Management SIG, ISMB 2009, Stockholm, June 2009.