Toward Efficient Variant Calling Inside Main-Memory Database Systems

Mutations in genomes indicate predisposition for diseases or effects on efficacy of drugs. A variant calling algorithm determines possible mutations in sample genomes. Afterwards, scientists have to decide about the impact of these mutations. Certainly, many different variant calling algorithms exist that generate different outputs due to different sequence alignments as input and parameterizations of variant calling algorithms. Thus, a combination of variant calling results is necessary to provide a more complete set of mutations than single algorithm runs can provide. Therefore, a system is required that facilitates the integration and parameterization of different variant calling algorithms and processing of different sequence alignments. Moreover, against the backdrop of ever increasing amounts of available genome sequencing data, such a system must provide matured database management capabilities to enable flexible and efficient analyses while keeping data consistent. In this paper, we present a first approach to integrate variant calling into a main-memory database management system that allows for calling variants via SQL.