Are Graph Databases Ready for Bioinformatics?

Many many months back, we looked into NoSQL databases for storing de Bruijn graph structure and mentioned about them here. Bioinformatics tools often use graph structures for storing various data sets. Some of those graphs are fairly large and need to be traversed quickly. Unlike traditional SQL databases, which are fairly slow for graphs, NoSQL databases like Neo4j seem to focus on rapid graph traversal as their main goal.

A new editorial article - “Are graph databases ready for bioinformatics?” - published in Bioinformatics by Christian Theil Have1 and Lars Juhl Jensen shines more light on the same topic.

Their conclusion -

In summary, graph databases themselves are ready for bioinformatics and can offer great speedups over relational databases on selected problems. The fact that a certain dataset is a graph, however, does not necessarily imply that a graph database is the best choice; it depends on the exact types of queries that need to be performed. Graph queries formulated in terms of paths can be concise and intuitive compared with equivalent SQL queries complicated by joins. Nevertheless, declarative graph query languages leave much to be desired, both feature-wise and performance-wise. Relational databases are a better choice when set operations are needed. Such operations are not as natural a fit to graph databases and have yet to make it into declarative graph database query languages. These languages are efficient for basic path traversal problems, but to realize the full benefits of using a graph database, it is presently necessary to tightly integrate the relevant algorithms with the graph database.

Speaking of de Bruijn graphs, our conclusion in early 2012 was that graph databases were too bulky for genome assembly.