Hypergraph Database Management System (HygDAS)

The current de facto standard of database management system is the relational model proposed by Edgar F. Codd in 1970 in his seminal paper entitled "A relational model of data for large shared data banks". It was well-accepted due to its relative ease of comprehension (tabular data structures) and query language implementation.

However, 4 decades on and the relational model exhibits a number of inadequacy for use in current day data management needs. For example, relational model cannot manage geo-spatial and multimedia data effectively. In the case of protein-protein interaction data extracted by text analysis using Muscorian, we are faced with a number of challenging issues - a protein/gene name may change over time due to renaming, splitting or merging (an example can be seen here). Storing and querying such data in a relational database system can be very complicated. Another example is genetic mapping data, essentially a geo-spatial data-type, which is also rather difficult to store and retrieve in a relational database system.

Gene name renaming/splitting/merging, as well as genetic maps, can be represented as a graph data-type. Hence, I think that a graph-based data management system may be suitable for bioinformatics needs. To improve the flexibility of the system, a hypergraph model is proposed. In a hypergraph, each node (known as a hypernode) can comprise of one or more nodes (simulating gene renaming and merging) and each edge (known as a hyperedge) or link can comprise of one or more edges.

At the same time, a relational model can be mapped onto a hypergraph model - entire table can be a hypernode, a record can be hypernode, a field can be a hypernode, and so on.