MemVerge™ announced that SeekGene has significantly reduced processing time and cost for data intensive single-cell analysis tasks using MemVerge Memory Machine running on AliCloud i4p compute instances. As a result, SeekGene is seeing a five-fold increase in output per virtual machine (VM) in its analytical operations. In particular, the data loading and exporting performance of its single cell sequencing pipeline has improved by two orders of magnitude, and it has doubled the sample size of the dataset used in the analyses.
SeekGene is a biotechnology enterprise focusing on single-cell technology that supports clinical diagnosis and development of precision medicine. The medical biopharmaceutical organization owns an exclusive microporous chip and water-in-oil dual technology platform and performs independent research and development of high throughput single-cell products, experiments, and full-chain services for bioinformatics analysis.
SeekGene’s SeekOne NGS single-cell library platform, the SeekGene Online automated online data analysis platform, and its proprietary droplet method and micropore method dual platform sequencing capabilities provide data analysis for international scientific researchers. The sequencing services are deployed on AliCloud. However, because analytical processes use expression data as high as hundreds of thousands of reads, sequencing analysis can fail on traditional VM instances due to insufficient memory. In addition, the export and loading process of temporary data on disk during sequencing tasks can also be extremely lengthy.
Using MemVerge Memory Machine Cloud Edition software running on AliCloud i4p VM instances, which feature Intel Optane persistent memory (PMem), SeekGene is now able to use large memory resources with no change to its code. This allows SeekGene to double its sample size and enables up to five times more concurrent processes to run. Further, with MemVerge Memory Machine Cloud Edition, SeekGene is able to improve data loading and exporting performance by two orders of magnitude by eliminating the I/O bottleneck that is caused by disk reads and writes. Specifically:
- On a traditional VM instance which uses NVMe SSD to save temporary data, it takes over 15 minutes to store the dataset. By employing MemVerge Memory Machine snapshot technology, saving the data takes only 2.5 seconds.
- When compared to AliCloud ESC.g5, the previous AliCloud VM instance used by SeekGene, the AliCloud ECS.i4p, together with MemVerge Memory Machine technology, enables SeekGene to run five concurrent tasks, each with twice the size of the original dataset.
“Using MemVerge Memory Machine, we are able to employ large memory resources in the cloud without refactoring our code, and eliminate the delay caused by storage I/O otherwise required in our pipelines,” said Xingyong Ma, Co-founder and Chief Scientist of SeekGene. “As a result, we are able to cut our analytical time and costs significantly while optimizing our single-cell sequencing capabilities for researchers worldwide to promote faster development of precision medicine.”
“The SeekGene use case is a typical example of how biotechnology researchers can revolutionize their computational analyses by leveraging Big Memory technology in the cloud,” said Jonathan Jiang, Chief Operating Officer, MemVerge. “These data intensive workloads can now be performed at record speeds and at dramatically lower cost. For the biotechnology industry, this can be a true gamechanger.”
MemVerge Memory Machine enables applications to utilize 100% of available memory capacity across multiple memory types with no code refactoring required, while providing new operational capabilities to memory-centric workloads. MemVerge Memory Machine Cloud Edition extends these benefits to cloud workloads, delivering memory virtualization, in-memory fault-tolerance and mobility services that organizations can easily add to their cloud infrastructure. Stateful, non-fault-tolerant, and long-running apps can now realize the promise of cloud agility and flexibility. More information on Memory Machine Cloud Edition is available here.
Image licensed by pixabay.com