Guest Post by Rami Mehio, VP of Engineering at Edico Genome
Today at AWS Summit San Francisco, Dr. Werner Vogels, Chief Technology Officer at Amazon.com, announced the general availability of their F1 instance. These EC2 F1 instances, Field Programmable Gate Array (FPGA)-enabled instances in the cloud, are top of the line. F1 Instances include extremely powerful Xilinx FPGAs. The launch of FPGA-enabled F1 instances on AWS means exciting updates for genomics cloud-users. The F1 instance enables us to further optimize DRAGEN to near real-time data analysis. It also allows us to deploy highly advanced and computationally intensive algorithms delivering unprecedented performance.
Along with the traditional use-cases that DRAGEN supports, DRAGEN on F1 can address the computational requirements of extreme high-coverage sequencing analysis, needed for applications such as somatic variant calling and structural variants at rapid speed. With the launch of DRAGEN on AWS, Edico Genome is also launching its just-in-time (JIT) compression framework for genomic data. JIT allows for on-the-fly, ultra-rapid storage compression and decompression of genomic data. JIT also offers a wide-variety of highly-efficient compression options.
DRAGEN runs on both types of instances offered by AWS. Customers can choose between the standard f1.2xlarge instance or the ultra-rapid f1.16xlarge instance. The F1.16xlarge instance offers massive FPGA-computational resources in the form of 8 VU9P FPGAs and 32 top of the line cores. The f1.16xlarge allows for the processing of 30x genome from FASTQ to VCF in ~15 minutes with room to further optimize to less than 10 minutes. DRAGEN on F1 instance types are highly scalable, available on-demand and incredibly accurate while offering the lowest possible cost solution on the market.
DRAGEN on AWS is currently only available for early access customers and includes all pipelines, just-in-time compressed data storage, and full automation using DRAGEN’s workflow management system. Edico Genome offers customers pay-as-you-go (per GB) or annual subscription pricing plans.
Also see AWS’ newest blog post about F1 Instances here: https://aws.amazon.com/blogs/aws/ec2-f1-instances-with-fpgas-now-generally-available/