Computational appliance: a supercomputer for modern biomedical research
Stanford University, Stanford CA
Investigators
Linked publications, trials & patents
Abstract
PAR16054 Application, PI: Somalee Datta, PhD, Stanford University NIH investigators at Stanford are increasingly analyzing terabyte to petabyte scale datasets generated using stateoftheart biomedical technologies. It is no longer unusual to find studies that analyze hundreds of samples, or correlate with other available large scale cohorts (e.g. UK10K, TCGA), or involve longitudinal multimodal data. Analysis and interpretation of these large scale complex data require a computational environment that is fast and affordable. Stanford researchers have access to computational capacity in the form of traditional clusters. These clusters exhibit significant IO bottlenecks thereby slowing down the rate and amount of analysis and in some cases, making the analysis impractical. Stanford has also made significant investment to prototype certain biomedical applications using the ?Big Data? distributed programming stack, Hadoop, such as memory intensive Spark clusters or Google BigQuery. Unfortunately, it is impractical to rewrite 100s of common biomedical applications for Hadoop. We therefore propose a supercomputer, that allows our biomedical applications to scale without additional investment in personnel to rewrite the commonly used tools. The supercomputer has large number of processors (256 CPU cores, 4 Tesla K80 GPUs) and memory (8 TB RAM, 16 TB ?NVMe flash?) to provide performance for large data. To optimally support the needs of our research community, we propose to colocate the supercomputer along with our existing computational capacity. The system will be hosted at a Stanford approved data center and managed by Stanford IT. Researchers using this device will have access to an extensive biomedical stack with over 350 commonly used tools as well as many commonly used reference datasets (e.g. 1000 genome) and access control datasets (e.g. GTEx or TCGA). With addition of the supercomputer, the computational environment will provide a full stack biomedical research environment including capacity, range of computational capabilities, extensive biomedical tools, common reference datasets, easily available training and consulting support. By making this supercomputer available to a large crossdisciplinary biomedical research community at Stanford, we expect to invigorate development of novel algorithms, mathematical and statistical approaches unhindered by the limitations of current capabilities found in typical academic clusters and public Clouds. Project Summary/Abstract
View original record on NIH RePORTER →