BioHDF-XML-RDF Triplet

The first products, based on BioHDF, will provide data models, APIs, software tools (I/O, algorithms), and a viewer based on HDFView, to support DNA polymorphism discovery and genotyping. Using BioHDF, researchers will be able perform resequencing-based SNP discovery, analyze genotyping data, and export datasets in formats ready for submission to key databases. As a programming environment, BioHDF will be easily extended to accept data from new genotyping platforms and format data for interchange with many databases. Additionally, BioHDF will be able to be used to support whole genome association studies and linkage disequilibrium (LD) calculations in very large data sets like HapMap. BioHDF will be delivered to the research community as an open source technology.

Does HDF5 (BioHDF) provide a better alternative for domain specific formats or it will be a complementary technology. I guess HDF5 alone can not support our requirements. Over the time people have realized that XML exclusively is not sufficient to represent the domain specific data objects, that means we need other frameworks such as RDF which is now extensively used for metadata associated with data objects. Next in this row is HDF5, and looking on the complexity of biological data objects it will be wiser to use a combined strategy incorporating XML, RDF and HDF5. Although BioHDF is targeting only genomic data, but in broader sense a technology based on HDF5-XML-RDF (or BioHDF-XML-RDF) triplet can map any type of biological object. XML will hold the data model, while use of RDF metadata framework can help to bring the context and provenance, and at the same time HDF5 will make it a scalable solution. HDF5 can hold metadata information too, but considering the RDF and its relationships with the Semantic Web, RDF framework will be certainly a better option. Basically BioHDF-XML-RDF Triplet look something like this-