How much compute power do you need to analyze 10 gene panels, exomes, or genomes each day? What about 1,000 genomes each week?

Questions about compute infrastructure are among the most common we hear from our customers, and this has been the primary motivator for our collaboration with Intel. Together, we’ve been working to support affordable, massively scalable whole-genome analysis based on Intel’s world-class infrastructure and our industry-leading analysis tools.

Earlier this year we released the reference architecture that came from this collaboration, and you can learn more about it in the webinar below featuring Michael McManus, Senior Health & Life Sciences Solution Architect at Intel, and Mikael Flensborg, Director and Solution Lead for QIAGEN Bioinformatics. The speakers will update you on the new architecture, which combines our Biomedical Genomics Solution with the Intel Scalable System Framework. They’ll also share advice on sizing and building a genomics cluster based on data-generation needs, as well as how to scale the cluster as demand changes.

The solution represents some of the code optimization work that Intel has focused on lately, teaming up with industry leaders such as QIAGEN Bioinformatics to improve algorithms based on the latest multi-core chip architecture. As genome analysis tools become increasingly complex — newer haplotype-aware callers, for instance, are far more computationally intense than older generations of variant callers — there is greater need for scientists to become experts in computational infrastructure. With solutions like the QIAGEN/Intel reference architecture, we alleviate that burden by recommending a scalable, easily managed system that allows users to focus instead on the science.

For example, this system meets the computational and analysis demands of Illumina’s HiSeq X Ten while providing whole-genome analysis for as little as $22 per genome. With Intel’s 32-node cluster, researchers can save as much as $1.3 million in total ownership costs compared to a vendor-recommended BWA+GATK variant calling pipeline run on an 85-node cluster.

Learn more about our High volume sequencing solution

Webinar

Reference Architectures for the QIAGEN Biomedical Genomics Solution

New resource available: Pairing QIAGEN Bioinformatics tools with Intel technology for scalable whole genome analysis

Scalable whole genome analysis

As we discussed in a blog post earlier, we’ve been working together with Intel to bring world-class infrastructure together with industry-leading genome analysis tools to enable massively scalable whole genome analysis at lower cost. Now, we have released a new white paper detailing the reference architecture and other technical information for our joint solution.

Designed to help NGS scientists keep their sequencing pipelines running smoothly even at capacity - all while saving money and producing better results - our solution provides whole genome analysis for as little as $22 per genome. It meets the computational and analysis demands of Illumina’s HiSeq X Ten, but Intel’s 32-node offering can save researchers up to $1.3 million in total ownership costs compared to the 85-node cluster recommended by the vendor for a BWA+GATK variant calling pipeline.

Here’s a quick look at what makes our solution different:

• Built-in analysis tools: The system uses Biomedical Genomics Server solution.
• Scalability: Designed to scale on-demand for computing, networking, and storage, the cluster allows labs to manage capacity easily and cost-effectively.
• Proven accuracy: While efficiency and cost-effectiveness is an important factor for NGS data analysis, accuracy in both variant calling and interpretation for the solution is proven to be among the best.
• User friendly: The solution masks the complexity of cluster computing with the easy-to-use Biomedical Genomics Workbench.
• Fast connection to data: We used a high-speed interconnect system based on Intel True Scale Fabric to link the compute nodes and centralized storage, providing up to 40 Gbps of bandwidth per port.
• Parallel storage: The solution incorporates Intel Enterprise Edition for Lustre, the world’s leading parallel storage system, to keep all the nodes, cores, and threads operating at high efficiency.

For more details, check out the full white paper.

Our tests showed that the 32-node system could process and analyze 48 genomes in 24 hours, on average - enough capacity to handle all the data produced by a HiSeq X Ten. We also tested the system with exome data and successfully analyzed approximately 1,440 human exomes every 24 hours.

Together with Intel we were presenting this joint solution at the Bio-IT World 2016 conference in a presentation addressing the growing demand for population-scale genomics.

If you’d like to learn more but are not able to attend the conference, please feel free to email us.
More information about Bio-IT World