Software and Data Analysis
Scan data originates from and is stored briefly on the MRI console. Additionally, as the scan session proceeds, scan images are sent to the Clinical Trials Processor (CTP) routing system on our main file server, where it is stored in our main DICOM repository. See the diagram below for the path taken by this data.
Clinical Trials Processor Work Flow at LCNI
Getting To Your Data
The repository for MRI data collected during scans is kept on our file server, Bordeaux. Access to data on this machine is provided through Windows networking (CIFS, aka SMB), SSH and SFTP. When your research group comes to the Center, an account will be set up for each member of your research team. These accounts allow you to log on to any workstation in the Center and also to the file server and compute grid. Once logged on, you will be able to work with your data.
Each individual account is generally associated with a lab account, and a link to this lab account directory will be created in the individual account home directory. Within your lab directory, a symbolic link named "scans-link" will point to your scan data in the central repository, which is rooted at /raid/data2/scans. The sub-directory structure there is organized as follows:
- lab directory - project ID - participant ID_YYYYMMDD
The bottom-level sub-directory holds all images from scan sessions for a participant performed on the date given by YYYYMMDD.
Certain departments associated with the University have a clear connection through our firewall using SSH. If you are using a computer in one of these departments, you will be able to connect to our file server and compute grid without further setup. If you receive no response from our machines, it is likely you are being blocked by the firewall. This will happen from off-campus locations and most departments on campus. In this case, you will need to use the LCNI VPN to establish a secure connection. Once the VPN connection is established, you will be able to log onto Center machines for data analysis and file management. File management may be done from the command-line or by using a graphical program that understands SFTP or Windows networking.
Our current recommendation for a graphical SFTP client is Filezilla, which is supported on Linux, OS X and Windows. Be sure to set the target server (Host field) as “sftp://bordeaux.uoregon.edu” or set the Port field to 22.
If you prefer the command prompt, run “sftp <username>@bordeaux.uoregon.edu”.
If you are using an OS X computer, you may connect to the server using Filezilla or through Terminal and SSH/SFTP.
Windows Networking (SMB/CIFS)
From a Windows XP machine, select the Start button, then Run... and enter “\\bordeaux” in the text field. On Windows 7 select the Start button and enter “\\bordeaux” in the text field. This will open an Explorer window into bordeaux. You will see the \public folder, which contains documents and programs that you may find useful, and your home folder, named with your account name. The lab's data storage area will be accessible through the "DICOM Central Repository" share. You may use the home folder and your lab folder as needed for analysis and storage of your data.
On OS X, networking connections are made using Finder. Select the Go menu, then Connect to Server (Command-K). In the server address box, enter “smb://bordeaux.uoregon.edu”. You will be prompted for your user name and password and then the shares you wish to mount (your home folder, Public, DICOM Central Repository, etc.). Once connected, the shares you choose will be available through the Finder.
Available machines for SSH connections are the file server (bordeaux.uoregon.edu) and the compute grid (ong<n>.uoregon.edu, where n is a number 0 through 11).
You are welcome to use the Center's compute resources for your processing. See Using the Oregon Neuroscience Grid for information on how to use this resource.
In addition to the resources at the LCNI, there are a number of other options available at the University and elsewhere. Center staff may be consulted about how to utilize these resources.
Local University Resources
The University offers data storage and High-Performance Computing solutions, available to all UO-affiliated research groups.
Applied Computational Instrument for Scientific Synthesis (ACISS) is the University's cloud computing system. It is by far the largest resource available locally, with a large number of compute cores, RAM and disk storage. It is equipped with many of the software programs typically used for fMRI data processing. If there are programs that you wish to use, but are not installed, we can have them installed for you.
To get started, visit https://aciss.uoregon.edu/newuser and enter as much information as you have available. This will be submitted to the ACISS administrators and you will be notified when your account is ready. You will then be able to log on to aciss.uoregon.edu. The initial log on screen is worth reviewing for detailed information on the ACISS facilities. It uses a "modules" system to load your environment with the necessary changes to use particular programs. To get the basics on using this system, log and type 'module help'. The command 'module avail' will list the software packages available. To use one of them, use the 'module load <package>' command. For instance, to use FSL, type 'module load fsl'.
You can even run interactive X11 programs.Once logged on to the head node, you use the qsub command to get an X11-tunneled connection to one of the many grid nodes.
From Chris Hoge:
After a bit of Googling, I found there's an easier way to do X forwarding directly from the scheduler.
You can ssh -X to the ACISS head node, then use:
qsub -q generic -l nodes=1:ppn=12 -I -X
The -q selects the generic queue, the -l nodes=1:ppn=12 guarantees that you reserve an entire machine, and the -X does the appropriate forwarding.
Extreme Science and Engineering Discovery Environment (XSEDE) is the largest set of resources publicly available for research computing. XSEDE consists of sixteen interconnected sites and offers terabytes of storage as well as petaflops of processing cycles. The University has a Campus Champion who can help your lab get started. Contact Sean Sharp for more information.