Transcription
Cryo-Electron Microscopy (CryoEM) is revolutionizing life sciences and pharmaceuticals by enabling atomic-level 3D visualization of proteins, viruses, and molecular complexes. CryoEM has become a critical tool for pharmaceutical and life sciences discovery, enabling researchers to develop new treatments and vaccines more quickly.
But for all of its potential, every CryoEM cycle generates massive, complex datasets – think 100TB or more of raw data comprising hundreds of thousands of individual files – that need to be moved to the cloud for analysis and imaging. Until recently, this required staging the data first to object storage in the cloud, then migrating it again to a separate high-performance NFS or Lustre service, where it can be read at high speed by an HPC cluster running CryoSparc, with the resulting high-resolution images being replicated back to on-premises storage.
For any given CryoEM cycle, not only would each successive replication job add a hundred or more terabytes’ worth of capacity usage but every file would also add transaction fees to the overall cost of every run. Further, the entire upload process can take days, and if there aren’t enough GPUs available in the target region for CryoSparc to run, the entire cycle either needs to wait until there are enough GPUs available, or the lab finds enough GPUs in a different region or cloud, and the whole process starts over. And once CryoSparc has written the output to cloud storage, replicating it back to the lab adds more time and costs to every run.
With Qumulo’s Cloud Data Fabric, powered by a Qumulo cluster in the main data center, a Qumulo Edge Accelerator appliance in the wet lab, and a Qumulo Cloud Accelerator instance in your choice of public cloud, your CryoEM workflow cycles can be shortened from days to hours. When the image files are uploaded from the microscope to the Edge Accelerator, they not only become immediately visible to clients connected to the on-premises Qumulo cluster, but also to the Qumulo Cloud Accelerator…
…meaning that CryoSparc can begin analyzing images from the cloud in real time as they’re ingested from the wet lab. And as CryoSparc generates new high-resolution, 3D images and writes them back to the Cloud Accelerator endpoint, they appear in real-time in the data center where research teams can begin using them.
With Cloud Data Fabric, there’s no need for complex, time-consuming replication to the cloud and no need for a separate, costly cloud-based file service. Qumulo Cloud Accelerators extend your on-premises file system to deliver data in real time and at line speed – reducing an hours-long process to literal seconds. And because a Cloud Accelerator can be fully functional on any public cloud in minutes without having to prestage data, your team can adapt in real time to changes in GPU availability across regions and clouds from one day to the next.
To learn more about Qumulo Cloud Data Fabric and Qumulo Accelerators, visit qumulo.com or contact your Qumulo representative today.