Biomedical Informatics is the interdisciplinary field that studies and pursues the effective uses of biomedical data, information, and knowledge for scientific inquiry, problem solving, and decision making, motivated by efforts to improve human health (AMIA).
At Sidra, our aim is to integrate these concepts using a cross-disciplinary approach combining the efforts of biologists, bio- and clinical information scientists, computer scientists, geneticists, mathematicians, software engineers and ultimately clinicians such as physicians and nurses.
We aim to create a world leading, innovative and cutting edge environment to foster research in Clinical Informatics and Bioinformatics.
Biomedical Research Informatics is not only a full-scale research division, it works in collaboration with Basic, Clinical and Translational Research. Our all-inclusive approach to work collaboratively with other research divisions within Sidra will help us to develop research studies that span beyond traditional Biomedical Informatics.
We will foster and develop strategic research collaborations with world leading universities and research institutes. We are undertaking cutting edge research in a number of topics and produce high impact publications to further our agenda of the being the most advanced Biomedical Informatics center making breakthrough discoveries that impact how healthcare is delivered.
We also envision creating a world leading IT service delivery model to support the IT needs for the entire research branch and also provide flexibility to our researchers to undertake research without constraints from the traditional IT environment. Given the nascent stages our development, our approach currently is to get our IT infrastructure up and running to support the research projects for the entire research branch. This infrastructure is a key to support genomic and biomedical research applications.
Biomedical informatics division within Sidra has acquired a state of the art multi-million dollar High Performance Computing (HPC) system specifically for Genomic and Biomedical Research applications
The architecture of Sidra HPC platform reflects specific challenges of Computational Genomics:
• Voluminous data
• Huge address space
• High dimensionality of data
• Non-deterministic Polynomial-time (NP) hard problems
• Poor parallel optimization
• Large number of applications, fast turnover of applications, mostly academic open source software
• Little or no floating point.
Sidra HPC platform (aka “Jernas”) is designed to answer these challenges. It has relatively low core density (640 per rack), large RAM per core, extra-large multi-tier GPFS storage and three cluster fabrics optimized for different types of data traffic.
The most demanding applications in Genomics pipelines (such as read mapping, fragment matching) are essentially variations of hash function that require large RAM. Few applications can scale to more than 256 cores. Our solution for data-intensive computation is based on virtualization of clusters of fat nodes to create multiple NUMA machines with large RAM and limited number of cores. We introduce rack-top QDR Infiniband specifically to optimize latency-critical data traffic between computing cores and RDMA. In benchmarks QDR Infiniband shows the best performance in tests with high volume of short messages.
The data passing between computing cores and storage is characterized by large volumes of relatively large files. This data required high bandwidth, but less sensitive to latency. For this purpose we introduce IBM TOR 10Gb Ethernet. In the future we plan to upgrade this fabric to 100GB Ethernet when hardware becomes available.
Bioinformatics consulting and collaborative research
We are focusing on setting our High Performance Computing environment that is being custom build for our needs.