A High Performance Storage Appliance for Genomic Data

Abstract

Rapid advancements in the area of next generation sequencing is revolutionizing the way in which biologists and now increasingly, clinicians analyze genomic data. These advances have substantially decreased the time and the cost it takes to sequence the genomes of new patients, thereby making genomic techniques more mainstream and giving rise to the new era of precision medicine. National scale genome programs have been launched in various parts of the world such as USA, the United Kingdom, and Saudi Arabia to name a few. One of the key insights out of this mainstream adoption is that even though the time and cost of generating sequence data has decreased dramatically, the cost of analyzing the data to yield clinically relevant information has not proportionally decreased. On the contrary, downstream analysis of the genomic data now dominates the cost in terms of time, effort and monetary value. This could be attributed to a number of factors: the sheer volume of data, limited knowledge of phenotypic, regulatory and epigenetic artifacts within the genome, and limited computational capabilities of existing data analysis tools and infrastructure. Overcoming these challenges is central to realize a more accurate, sophisticated and cost-effective genomic medicine. Another challenge, related to the limited analytic capabilities of existing computational and storage infrastructure is what we address in this paper. We discuss how novel trends in hardware, including the emergence of cheap, high performance and endurance solid-state storage associated with low latency interconnect and software defined orchestration, can help creating a high performance storage tier which improves data acquisition, storage, transmission and analysis over the current commercial alternatives.