FYI: Are graph databases ready for bioinformatics?

[Michael Hunger]

http://bioinformatics.oxfordjournals.org/content/29/24/3107.full

Neighbor network	Best-scoring path	Shortest path
PostgreSQL	206.31 s	1147.74 s	976.22 s
Neo4j	5.68 sa	1.17 s	0.40 s
Speedup	36×	981×	2441×

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.

[Viktor Bojović]

Hi Michael,

I didn't read the paper and don't understand how neo4j MATCH keyword works. Does it join by integer or string? If in left example "Interactions" is postgresql table, then join is performed by varchar which is very slow, and should be replaced with integers. Then , I believe, performance test would be more adequate.

--
You received this message because you are subscribed to the Google Groups "neo4j-biotech" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neo4j-biotec...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
---------------------------------------
Viktor Bojović
---------------------------------------
Wherever I go, Murphy goes with me