Evaluating FlexPod for genomics applications

Last month I was off work for 12 days. It began as a trivial wound on my foot, but rapidly developed into cellulitis and needed surgical intervention. The doctors sent the tissue sample to the lab to identify the aggressive bacteria responsible for the infection and to identify an antibiotic suitable for killing them. But in the interim, they had to start an empirically reliable, broad-spectrum antibiotic because the lab needed 3 to 4 days to culture the sample and verify the antibiotics capable of tackling the infection. As I was lying in a hospital bed with my foot elevated, I wondered how technology could reduce the wait time of 3 to 4 days and help patients receive a reliable antibiotic sooner.

As it happens, it’s possible to genetically analyze the sample to identify the infection agents and the antibiotics against which they have (and have not) developed resistance. So genomics has already solved this problem—in theory. However, there are some practical challenges before the latest advances in genomics can be applied to improve quality of care. A clinical genomics solution needs to mitigate challenges such as:

• IT infrastructure should be able to cost effectively support heavy compute and storage requirements.
• Researchers should be able to share data safely and seamlessly across multiple premises and multiple clouds.
• Data scientists should get a turnkey AI solution with full stack for training and using artificial intelligence models, without IT admins having to figure out how to configure the AI stack.

Earlier this year, we published an infographic about how FlexPod can address the infrastructural challenges for genomics and can help patients receive benefits from the advances in genomics. I had also blogged about how the FlexPod® converged infrastructure solution from Cisco and NetApp can serve as the IT backbone of genomics workloads. 

Now FlexPod is going to publish a proof-point in the form of a technical report for genomics applications.

In this TR, we will validate a FlexPod solution for deploying GATK (Genomics Analysis Tool Kit by Broad Institute), the most popular suite of applications for executing genomics analysis pipelines. The report will also offer guidance for optionally deploying NVIDIA GPUs as part of a FlexPod converged infrastructure. That infrastructure will provide superfast performance for any genomics application that requires parallel processing, or any deep learning AI model that needs to be trained fast using the GPUs.

This technical report will make it easy to set up FlexPod for end-to-end processing of genomic workloads. It also discusses how FlexPod supports genomics use cases by offering:

• A common storage platform across on premises and multiple private and public clouds, so that genomics data can be moved to any location as necessary for sharing with researchers around the globe
• NetApp® ONTAP® FlexGroup for handling the billions of files that are generated in next-generation sequencing
• Cost-effective auto-tiering and archiving necessary to support the large storage requirements of genomics use cases
• Deduplication and compression efficiencies that help to reduce resource utilization
• End-to-end NVMe for faster access to genomics data