Neuroscience Informatics

Senior Scientist: Dr. Neil Killeen
Group Members: Wei (Wilson) Liu, Sina Sadeghi
Collaborators: Jason Lohrey (arcitecta), Steve Melnikoff (Victorian eResearch Strategic Initiative - VeRSI)

Collaborative Neuroscience: Neuroinformatics for Sharing Data and Tools

One major challenge of the 21st century is the understanding of the human brain, the most complex organ created during evolution. The new interdisciplinary field of Neuroscience Informatics (NI) capitalizes on the potential synergies between neuroscience and information technology. A multinational working group on NI has been set up by the Organisation for Economic Co-operation and Development (OECD) to address some of the challenges facing efforts to achieve a truly global collaborative neuroscience. The OECD Neuroinformatics Working Group was established in 2000 with the objectives of (i) developing a common resource for neuroinformatics tools and databases, (ii) establishing guidelines and recommendations for their organization and interoperability, and (iii) assisting with the communication and dissemination of worldwide efforts in neuroscience informatics.

One significant outcome from the OECD Neuroinformatics committee was the creation of the The International Neuroinformatics Coordinating Facility. It's mission is to coordinate and foster international activities in Neuroinformatics through workshops, tools and standards. Although Australia is not an official signatory to the INCF, the Australian Neuroinformatics community participates in INCF activities.

Within the Australian community, Neuroimaging projects are now becoming truly National in their scope. For example, the NCRIS National (Animal) Imaging Facility will have nodes distributed throughout Australia. Similarly, the Australian Mouse Brain Mapping Consortium (NHMRC) will generate data in Queensland, Victoria and New South Wales and have multiple, geographically dispersed nodes that will share the data.

Recognizing that the challenges of such projects lie not only in the computational throughput, but also in the management of information, the Neuroinformatics team are developing a Neuroimaging Informatics and Computational Facility. This Facility, seeded by the ARC, is being developed and supported jointly by the Centre for Neuroscience (University of Melbourne), the Howard Florey Institute and the Department of Electrical and Electronic Engineering (University of Melbourne). The Facility has two main streams of activity at present:

  • Providing an advanced computational and data storage platform
  • Developing a distributed Neuroimaging data management system

Computational Platform

  • An SGI Altix share-memory 3700 Bx2 system with 64 1.5GHz Itanium II processors
  • Each processor has 4Gb of local memory. However, all processors can share all of the memory (256Gb)
  • A single Linux image manages all of the processors
  • Data storage if 4.8TB of disk and 57Tb of tape
  • The tape storage is provided by an extensible StorageTek SL500 Tape Library which operates with three LTO3 tape drives and one robotic controller
  • Data migrate between disk and tape (together the Ôdata storeÕ) in a process controlled by the Data Migration Facility (DMF) software
  • Data have redundancy through the disk RAID configuration and two copies on tape
  • There is no external copy of the data store at this time

Data Management
The team is developing an informatics system that receives, organizes and distributes Neuroimaging information. This system is built with Mediaflux, a commercial informatics platform available from Arcitecta Current capabilities are:

  • Web portal-based (https) username and password authentication
  • Role-based authorization so that data are accessible only to authorized users
  • Receive and manage MRI/DICOM data from the RCH and HFI's aMRIF
  • Data access (query and download) via web-based portal

New development will modify the system to be more generic; it will handle other data acquisition types other than MRI/DICOM such as Histological data, handle more complex workflows, and we will also develop the capability to federate with other systems.