Intro to NITRC Computational Environment

[Arman Eshaghi]

Apologies for the spam, I thought Python gurus might be interested in NITRC on cloud. 

FYI - feel free to spread the word.

Webinar: Fall 2014, date TBA

Topic: Intro to NITRC Computational Environment (NITRC-CE)

Learn about the different ways you can use the NITRC Computational Environment, see a live demo, and have an opportunity to question the developers first hand, and talk about any challenges you've had to date so we can fix them.  

You can use NITRC-CE on your own infrastructure or use it on a commercial cloud provider either via AWS Marketplace, a public AMI Interface, or StarCluster Interface with Grid Engine (coming soon to Microsoft VM Depot as well).  

According to the University of Liège, NITRC-CE helped "Reduce time processing neuroimaging data by 85% allow[ing] me to complete a critical stage of my research in 2 days, instead of 2 weeks." 

For more information see: http://www.nitrc.org/projects/nitrc_es/

To sign up for the event, please email nina....@tcg.com  Up to 10 participants will be signed up and webinar conference log-in will be provided at that time.

[Erik Ziegler]

For what it's worth, the ULg quote was related to our recent paper on fiber track density in Parkinson's disease (http://www.ncbi.nlm.nih.gov/pubmed/24956065).

The Nipype pipelines (using Dipy, ANTs, FSL, and MRtrix) can be found here:

https://github.com/CyclotronResearchCentre/parktdi_scripts

All the imaging data can be found here:

https://www.nitrc.org/projects/parktdi/

They're quite messy at the moment but the track normalization may be made into Nipype workflow later. At the time I was running two instances for different pipelines, so I had some code to send an email with the instance ID when a pipeline finished:

https://github.com/swederik/AmazonEC2scripts

I would definitely recommend trying NITRC-CE out (specifically, the Cluster Compute version: https://aws.amazon.com/marketplace/pp/B00DLI6VAQ/ref=sp_mpg_product_title?ie=UTF8&sr=0-3) if you have a lot of data to process. It works flawlessly with Nipype pipelines and the learning curve is not particularly steep. If you've just signed up, you'll be able to learn the ropes using the free micro instances. I would recommend attaching an EBS-backed volume initially if you plan to keep your data. It's a big hassle to move large amounts of data from ephemeral storage later on.

For anyone using MRtrix on the cluster compute instance: make sure you set the maximum number of cores to 4 or less. The 32-core instances are built from 8 machines and in my experience the MRtrix multi-core processing fails to distribute properly across all cores. If I had spent a little more time setting up EC2 and StarCluster, I'm sure I could have processed all 53 subjects in about 8-10 hours.

If anyone has any questions I'm happy to help,

Erik

--

---
You received this message because you are subscribed to the Google Groups "NiPy Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to nipy-user+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.