The GMQL system is freely available for non-commercial use as open source project at: http://www.bioinformatics.deib.polimi.it/GMQLsystem/.

In previous work, we presented GenoMetric Query Language (GMQL), an algebraic language for querying genomic datasets, supported by Genomic Data Management System (GDMS), an open-source big data engine implemented on top of Apache Spark. GMQL datasets are represented as genomic regions (i.e. intervals of the genome, included within a start and stop position) with an associated value, representing the signal associated to that region (the most typical signals represent gene expressions, peaks of expressions, and variants relative to a reference genome.) GMQL can process queries over billions of regions, organized within distinct datasets. In this paper, we focus on the efficient execution of regionpreserving GMQL operations, in which the regions of the result are a subset of the regions of one of the operands; most GMQL operations are region-preserving. Chains of region-preserving operations can be efficiently executed by taking advantage of an array-based data organization, where region management can be separated from value management. We discuss this optimization in the context of the current GDMS system which has a row-based (relational) organization, and therefore requires dynamic data transformations. A similar approach applies to other application domains with interval-based data organization. Index Terms-Big data processing, data management, cloud computing, genomic computing.

With the huge growth of genomic data, exposing multiple heterogeneous features of genomic regions for millions of individuals, we increasingly need to support domain-specific query languages and knowledge extraction operations, capable of aggregating and comparing trillions of regions arbitrarily positioned on the human genome. While row-based models for regions can be effectively used as a basis for cloud-based implementations, in previous work we have shown that the array-based model is effective in supporting the class of regionpreserving operations, i.e. operations which do not create any new region but rather compose existing ones. In this paper, we remove the above constraint, and describe an array-based implementation which applies to unrestricted region operations, as required by the Genometric Query Language. Specifically, we define a wide spectrum of operations over datasets which are represented using arrays, and we show that the arraybased implementation scales well upon Spark, also thanks to a data representation which is effectively used for supporting machine learning. Our benchmark, which uses an independent, pre-existing collection of queries, shows that in many cases the novel array-based implementation significantly improves the performance of the row-based implementation.