Comparing relational and graph databases for pedigree data sets

Increasingly large family pedigree data sets are being constructed from routine civil and religious registration data in various parts of the world. These are then being used in health, social and genetic research in a variety of different ways. Often the type of questions that are being asked involve complex queries, such as the degree of relatedness between multiple sets of individuals, and involve traversing through the data typically multiple times. There is therefore an important issue of efficiency of querying. In this paper we evaluate the suitability of two classes of database, relational (MariaDB) and graph (Neo4j), for storing and querying pedigree datasets representing millions of individuals. We report results of measurements of scalability, query performance, and the ease of query expression, using synthetic datasets.