HBM Hackathon News

Contest Winners Announced!

NITRC ADMIN — Wed, 26 Jun 2013 12:39:00 GMT

Congratulations to the winners of the HBM Hackathon Competition!

Challenge 1: No Sound Consensus

Challenge 2: SCI

Challenge 3: MindBoggle-102

The winners of the HBM Hackathon Contest brought together multi-disciplinary teams to deliver excellent responses to the three challenges. Below is a list of the winning teams and their members for each category along with the submitted abstracts, presentations, and demo links. Join me in congratulating our winners and enjoy!

Challenge 1. Best imaging and gene expression relationship discovered via integration of imaging data with the Allen Human Brain Atlas.

Team “No Sound Consensus” - Rhodri Cusack, Mark Daley, Charlotte Herzmann, Annika Linke, Jonathan Peelle, Conor Wild, Leire Zubiaurre-Elorza

Presentation Link

Visualization Demo

Abstract:

There is little consensus on the parcellation of human auditory cortex, especially in regards to auditory-related or auditory association areas beyond primary auditory cortex; in contrast, there is substantial agreement on the functional parcellation of visual cortex. This might be because: fewer scientists study the auditory system and the key analyses are still lacking; auditory regions are smaller and/or too variable across individuals; or because the auditory system is fundamentally less modular, perhaps because the statistics of environmental sounds are such that processing is best implemented in a distributed monolithic system. Our hackathon project had two overarching goals: (1) to quantify modularity in auditory and visual systems; and (2) if appropriate, to derive a parcellation of the auditory system. As any single type of data is subject to biases, three types were brought together by an international and multidisciplinary team of scientists using multiple packages and programming languages. We first defined broad auditory and visual seed regions of interest using an open-source meta-analytic approach (i.e., http://neurosynth.org/) combined with anatomical masking. We then characterized the signature of connectivity of each voxel in these seed regions using diffusion and resting state data provided by the Human Connectome Project. Graph theory analyses were then applied to derive clusters of voxels (i.e., modules) that exhibited similar patterns of anatomical and functional connectivity, and to compare the modularity of auditory and visual cortices. Modularity of auditory cortex was found to be similar to that of visual cortex, suggesting that across individuals this system appears to have multiple distinct sub-regions. Parcellations in all seed regions showed consistency across individuals and modalities, and allowed us to derive group and grand (i.e., multi-modal) parcellations. Finally, using the Allen Brain Atlas we found that gene expression was more similar within our parcellations for visual – but not auditory – cortex. In summary, auditory cortex was found to be modular and to show consistency across individuals, so that reliable group parcellations were be derived. Finally, we developed a web tool that can be used to browse our parcellations and the connectivity of each module. We estimate that our entry has used around 1 year’s worth of processing time of the fastest processing cores in the Amazon cloud, and that much of team was sleep deprived for the last month.

Challenge 2. Best neural systems model or visualization based on large-scale integration of resting state fMRI data with other HBM Hackathon accessible datasets.

Team “SCI” - Miriah Meyer, Tom Fletcher, Sam Gerber, Sean McKenna, Wei Liu, Brandon Zielinski, Kris Zygmunt

Presentation Link

Abstract:

I will be presenting a tool that enables the visualization and exploration of the gene expression data in both the correlation space and MNI space. By navigating and visualizing in both spaces concurrently, the gene expression data can be explored and understood more completely, helping to generate hypotheses for further testing.

Challenge 3. MNI Mash-up: Most innovative map or aggregation of information in the MNI 152 standard.

Team “MindBoggle-102” - Arno Klein, Jason Tourvillle, Jay Bohland, Rich Stoner

Presentation Link

Visualization Demo

Abstract:

Provide anatomical labels, sulcal features, and shape information for 101 brains in MNI152 space.

Manually label, extract features, and analyze the shape of the cortex of an Allen human brain.

Create a browser-based application to display region-specific gene expression data.

Mash-up components:

Allen Human Brain T1 image and gene expression data

DKT cortical labeling protocol

Mindboggle-101 set of brains manually labeled with the DKT protocol

Mindboggle software for feature extraction, identification, and shape analysis

Web framework for visualization

MNI152 template

Yesterday, we extracted and computed shape measures on all 101 brains in the Mindboggle-101 data set, and are doing the same for the Allen brain right now. Today, we propose to manually label the Allen brain using the DKT protocol and create a browser-based application to enable a user to select a labeled region in the Allen brain and display gene expression data for that region.

Congratulations!

INCF sponsors the HBM Hackathon

NITRC ADMIN — Tue, 18 Jun 2013 3:48:00 GMT

The International Neuroinformatics Coordinating Facility (INCF) coordinates collaborative informatics infrastructure for neuroscience.

INCF is an international organization launched in 2005, following a proposal from the Global Science Forum of the OECD to establish international coordination and collaborative informatics infrastructure for neuroscience – and currently has 17 member countries across North America, Europe, Australia and Asia. INCF establishes and operates scientific programs to develop standards for neuroscience data sharing, analysis, modeling and simulation while coordinating an informatics infrastructure designed to enable the integration of neuroscience data and knowledge worldwide and catalyze insights into brain function in health and disease.

Mindboggle 101: Manually labeled brain surfaces and volumes

NITRC ADMIN — Tue, 18 Jun 2013 12:03:00 GMT

Mindboggle-101: Manually labeled brain surfaces and volumes

The Mindboggle-101 dataset includes manually labeled anatomical regions for 101 healthy subjects. The manually edited cortical labels follow sulcus landmarks according to the Desikan-Killiany-Tourville (DKT) protocol. The protocol, individually labeled brain images, optimal average surface and volume templates, and a surface Gaussian classifier atlas are all available for download and are described in the following article:

“101 labeled brain images and a consistent human cortical labeling protocol”
Arno Klein, Jason Tourville. Frontiers in Brain Imaging Methods. 6:171. DOI: 10.3389/fnins.2012.00171

The 3rd HBM Hackathon challenge: Resources Mapped to MNI Space

NITRC ADMIN — Mon, 17 Jun 2013 2:10:00 GMT

MNI Mashup: Most innovative map or aggregation of information in the MNI 152 standard.

This challenge challenge is to be addressed entirely during the meeting using resources made openly available to the brain mapping community.

Example projects:

Aggregating standard MNI space 3D shape models of neuroanatomical structures

“Dense brain information map”. Given an (x,y,z,r) tell me everything you can about (x,y,z,r), especially including uncertainty of that information. Here are a few examples of information that could be returned:
- gene expression (and the uncertainty would be high if imputing and low if exactly at a probe location)
- cognitive atlas terms and therefore publications and derivatives associated with that location ( a la neurosynth)
- variation in cortical thickness
- functional connectivity
- likelihood of major fiber bundles in that location
- registration inconsistency/uncertainty
- cytoarchitecture variation (neuron types, etc.,.)
- associated disorders
- data from other species

Judging criteria:

Scientific impact

Commitment to open availability of the resulting resource

Extensibility

Inclusion of quantified uncertainty

Rules

Participants must use publicly available data that is listed on the HBM Hackathon Blog. If a public dataset is not listed that you want to use, we would love to add it to the list, just contact us: hbm.hackathon@gmail.com

Participants can use any computational resources available to them, but judging will take into account innovative use of cloud computing and how openly available the approach is (see Judging)

At least one team leader/presenter must attend the meeting

Off-site team members are allowed and encouraged, but will not be eligible for all resources made available to registered OHBM attendees (e.g., cloud computing credits)

NeuroDebian: The ultimate neuroscience software platform

NITRC ADMIN — Wed, 12 Jun 2013 1:00:00 GMT

NeuroDebian is a turnkey software platform for nearly all aspects of the neuroscientific research process.

NeuroDebian provides a large collection of popular neuroscience research software for the Debian operating system as well as Ubuntu and other derivatives. Moreover it provides some popular datasets and atlases within the same convenient package management system.

Software

Popular neuroscience-oriented packages include AFNI, FSL, PyMVPA and many others. In addition popular projects for generic (e.g. pandas, xppaut) and distributed (e.g., coop-computing-tools, condor) computation needs are included. The entire list is too long to cite here.

Data

Atlases provided by software such as FSL and AFNI

NeuroSynth dataset and its FSL atlas

Haxby 2001 dataset

PyMVPA tutorial (including data and IPython notebook)

See our website for the exhaustive list

Installation

Majority of software packaged by NeuroDebian team is integrated are already available on any stock Debian or Ubuntu system. In addition we maintain a dedicated NeuroDebian repository which you could add to APT sources on your stock Debian/Ubuntu installation to obtain most recent versions and software we maintain AND data packages which are absent from official Debian and Ubuntu archives.

Personal virtualization

Pre-crafted NeuroDebian virtual appliance allows to start using NeuroDebian and thousands of available software packages on any operating system in matter of minutes.

Cloud

NeuroDebian provides software for the NITRC-CE environment available for the hackathon participants.

Support

Email us directly with any “private” communication. Otherwise please use our public mailing list with any support questions.

You are welcome also to join #neurodebian IRC room on OFTC network if you have quick questions or want to join a live discussion.

Please visit our support page for additional channels.

Conclusion

If you are not using NeuroDebian already - you must try it now!

NIF: an inventory of Web-based neuroscience resources

NITRC ADMIN — Wed, 29 May 2013 9:03:00 GMT

The Neuroscience Information Framework: neuroscience data, materials, and tools via any computer connected to the Internet.

Background

Currently NIF Offers:

A search portal for researchers looking for neuroscience information, tools, data or materials

Access to content normally not indexed by search engines (i.e, the “hidden web”)

Tools for promoting interoperability among databases

The NIFSTD ontology covering the major domains of neuroscience (e.g., brain anatomy, cells, organisms, diseases, techniques)

Developer resources for accessing the NIF vocabulary and NIF tools, such as web services and including a tutorial

Some particular data sets which can be offered are Brain Connectivity Data (50316 records), Brain Activation Foci (5,145), Diseases (1,521), Microarray (642,995):

Integrated Nervous System Connectivity View is an aggregated dataset of connectivity statements from BAMS, CoCoMac, BrainMaps, Connectome Wiki, the Hippocampal-Parahippocampal Table of Temporal-Lobe.com and the Avian Brain Circuitry Database.

Integrated Disease View is a virtual database currently indexing authoritative information on disease and treatment options from: NINDS Disorder List and PubMed Health.

Activation Foci data provides functional activation foci data from published neuroimaging studies.

Example Responses to Challenges

Challenge 1:

Use NIF neuron registry to identify cell systems in various neurodegenerative diseases and map it to ABA data.

For different diseases, use NeuronRegistry-NeuroLex to infer from genes what specific cell populations differently effect neuron types. Also, we’d like to link those cell systems in Neurodegenerative diseases to neurotransmitter data in NeuronRegistry.

For example, in Alzheimer we want to examine ABA genes and how differently they affect different group of neurons found in NeuronRegistry. Infer what cell populations impacted in various brain regions.

ABA has mapped neurotransmitter related genes, thus one challenge is: Are there specific neurotransmitter systems affected in mental disorders and do they map to any known cell type?

Challenge 2:

Use NIF cell data as a bridge between resting data and projections of know cell types in various brain regions (ABA). Use Neurolex (NIF) and tractography (NIF) data we can estimate what type of cells where can cell body are and where axon projected.

Explore and get public fMRI datasets from COINS Data Exchange

NITRC ADMIN — Tue, 14 May 2013 8:29:00 GMT

COINS manages fMRI, MEG and phenotypic data for over 400 studies across six research institutions.

An ever-increasing subset of this data is being made publicly available on the COINS Data Exchange. Currently, the research community can explore and request raw MRI, fMRI and phenotypic data from several independent studies.

Tutorial:

A simple video-tutorial for querying can be seen on the COINS Data Exchange site, or on YouTube.

Exploration:

Creating a COINS account is easy. Just go to http://coins.mrn.org/dx, enter your email address, make up a password and click ‘Get Account’. That account is needed to let you know when your data are available for download, and to track any data agreements necessary for specific datasets. It is recommended that you do this prior to the Hackathon, just to explore the Data Exchange before needing to use it.

Once your account has been created, you can browse data in the Data Exchange. A unique filtering tool allows you to construct complex queries by combining data filters into logical groups and sub-groups. The goal of the exploration should be to target the exact data that is useful to you before you request and download it. This will save bandwidth and computing time on the COINS servers. Additionally, a well-targeted request will save you time by decreasing the need to post-filter data after download.

Approval:

Once you have found your ideal data, you can request that data by clicking the ‘Request’ button at the top-right. Though all data in the COINS data exchange is fully anonymized and defaced, some studies will need to explicitly approve your request. For this reason, you can provide information about your intentions for the data that you are requesting. Other studies (ABIDE, for example) have pre-approved their data for instant download. Each data-source is approved independently, so you can start downloading data as it is approved.

Delivery:

Approved data is packaged into ‘Data Capsules’ on the COINS servers. You will receive an email when your Data Capsule is ready, and you can then begin your download(s).

Using COINS Data Exchange to meet your OHBM Hackathon objectives:

At present, the COINS Data Exchange is only accessible through the browser interface at http://coins.mrn.org/dx. However, once your data has been approved and packaged for download, you can download each data capsule directly to another web-server or cloud-based storage device from the command line or a script.

After choosing which capsule to download, a dialog will appear with a temporary URL that can be used to request the capsule directly. The link will expire as soon as the download button is clicked, or after five minutes, so please do not click on the data until you are ready to download it..

Downloaded imaging data may consist of DICOM or NIfTI files, and will be organized in the following file-hierarchy:

Downloaded phenotypic (assessment) data will be contained in CSV files: one per instrument (questionnaire form).

Currently Available Datasets:

ABIDE: from http://fcon_1000.projects.nitrc.org/indi/abide/: In response, the Autism Brain Imaging Data Exchange (ABIDE) hereby provides previously collected resting state functional magnetic resonance imaging (R-fMRI) datasets from 539 individuals with ASD and 573 typical controls for the purpose of data sharing in the broader scientific community. Various cognitive and clinical assessments are also available. These data are pre-approved for sharing.

Discovery Sci: http://fcon_1000.projects.nitrc.org/indi/pro/nki.html: The NKI/Rockland Sample is intended to be a phenotypically rich neuroimaging sample, consisting of data obtained from individuals between the ages of 4 and 85 year-old. All individuals to be included in the sample undergo semi-structured diagnostic psychiatric interviews, and complete a battery of psychiatric, cognitive and behavioral assessments in order to provide comprehensive phenotypic information for the purpose of exploring brain/behavior relationships. Data from the NKI/Rockland sample will be given away prospectively, during the course of acquisition (randomized 2-8 week lag). These data require a data sharing agreement which can be quickly approved.

MCIC: The MIND Clinical Imaging Consortium is composed of The Mental Illness and Neuroscience Discovery Institute, now the Mind Research Network (MRN, http://mrn.org), comprised of investigators at the University of New Mexico, the University of Minnesota, Massachusetts General Hospital, and the University of Iowa (data from this site not available). The MCIC consortium conducted a cross-sectional study to identify quantitative neuroimaging biomarkers of schizophrenia. Expertly collected, well-curated data sets consisting of comprehensive clinical characterization and raw structural, functional and diffusion-weighted DICOM images in schizophrenia patients (n=162) and sex and age-matched controls (n=169) are now accessible to the scientific community. These data also require a data sharing agreement which can be quickly approved.

Contact:

The COINS developers are excited to support the OHBM Hackathon participants. Any requests, questions or feedback can be directed to ni@mrn.org.

NITRC-CE: A computational resource for the HBM Hackathon

NITRC ADMIN — Mon, 13 May 2013 7:00:00 GMT

NITRC-CE is an on-demand, cloud based computational virtual machine pre-installed with popular NITRC neuroimaging tools.

In January of 2013, the Neuroimaging Informatics Tools and Resources Clearinghouse (NITRC) released it’s Computational Environment (NITRC-CE), an on-demand, cloud based computational virtual machine pre-installed with popular NITRC neuroimaging tools built using NeuroDebian.

Purpose

The NITRC-CE is designed to be a low-barrier entry point to accessing the power of the cloud-based computing environment by providing a fully configured, ready to go EC2 instance within just a few clicks of the mouse.

Access

NITRC-CE is available via the Amazon Marketplace. In addition, you can also access a ‘public AMI’ to conduct your analyses on the Amazon EC2 platform. The NITRC-CE is scalable and extensible, and preconfigured with many software packages. Additional software supported under Ubuntu 12.04 LTE can be added by the user.

Data

Each NITRC-CE instance has remote access, so that users own data can be copied to and from using ‘scp’ and ‘sftp’. In addition, The NITRC-CE desktop is one-click away from the NITRC Image Repository (NITRC-IR), home of data for the 1000 Functional Connectomes, ADHD-200, ABIDE and other datasets. In addition, the NDAR download manager is available to assist in access to your NDAR datasets. Additional access points to other Hackathon data will be added in the near future.

Support

The staff of NITRC are available to assist users with issues (nitricinfo@nitrc.org), and a comprehensive Users Guide and discussion forum dedicated to HBM Hackathon support.

Conclusion

Best of luck in your participation in the HBM Hackathon. We hope the NITRC-CE will be a valuable resource to you. Feel free to contact us with any questions, problems or suggestions for enhanced support of this endeavor.

Open Brain Mapping at the 1st HBM Hackathon

David Kennedy — Mon, 06 May 2013 6:51:17 GMT

OHBM 2013 is only six weeks away, bringing the brain mapping community to Seattle, a city at the forefront of open neuroscience and information technology. The Local Organizing Committee, the Allen Institute for Brain Science, and Amazon Web Services are excited for the opportunity to highlight these aspects of our field in the HBM Hackathon (http://humanbrainmapping.org/hackathon), a meeting-long event integrated with and running parallel to the activity in the poster and exhibition space.

Our goals are to accelerate the development of a critical mass of cloud-based data, analytic, and computational resources for human brain mapping, and to provide OHBM attendees with access to and knowledge about them.

The event is being supported by a broad spectrum of partners from the open neuroimaging community, making it possible to allow an unprecedented degree of access to key resources in the field, including the Allen Human Brain Atlas, the first data release of the Human Connectome Project, the NITRC Computational Environment, and more. Experts from the Allen Institute, Amazon Web Services, Human Connectome Project, NITRC, INCF, LONI, NDAR, BIRN, and others will bring their expertise on-venue.

Register for the HBM Hackathon (http://humanbrainmapping.org/hackathon) to participate in a contest organized around three challenges, to tinker less formally, or just to be open to open neuroscience. You’ll also receive $100 in free AWS credits to make use of cloud computing technologies.

The contest is organized around three Challenges:

Challenge 1. Best imaging and gene expression relationship discovered via integration of imaging data with the Allen Human Brain Atlas. 

Challenge 2. Best cortical neural systems relationship or visualization based on integration of resting state fMRI data with other HBM Hackathon accessible datasets.

Challenge 3. To be announced the week of the meeting and pursued completely on site.

Winning entries in each category receive:

An invitation to submit project to Frontiers in Brain Imaging Methods with Open Access Publication Fees waived (must undergo standard peer review)
Amazon Kindle Fire and/or Paperwhite (limit 3 per team)
AWS hosts AMI and/or data resulting from the hackathon free of charge
Free GitHub Membership with private repositories (Challenge 1: one year silver, Challenge 2: 1 year bronze, Challenge 3: six months bronze)

To learn more, visit the HBM Hackathon Blog (http://ohbm-seattle.github.io) or join the discussion group (http://www.linkedin.com/groups/HBM-Hackathon-4957800).

And be sure to register at: http://humanbrainmapping.org/hackathon

Human Connectome Project Q1 Release: Now Cloud Accessible

NITRC ADMIN — Fri, 03 May 2013 8:00:00 GMT

The first full quarterly HCP Data release is now Cloud accessible for participants who have registered and agreed to the Open Access Data Use Terms.

HBM Hackathon participants will have the opportunity for high speed access to the first full quarterly HCP Data release via Amazon Web Services. All data available from the HCP Connectome-in-a-Box can now be downloaded directly from S3, Amazon’s Simple Storage Service.

This post is organized into three sections:

Description of the HCP Q1 Release

Getting Authorized to Access the HCP Data

Accessing HCP Data on AWS

Description of the HCP Q1 Release

The Q1 data release consists of multimodal MRI data collected from 68 healthy young adults who were scanned in the fall of 2012. These include all 12 subjects from our Initial Data Release. All 3T MRI scan data is included: Structural, Functional (resting state and task) and Diffusion. Behavioral data collected on all subjects is also included, with the exception of sensitive restricted-access data. Please see the Full Release Documentation for further details.

Getting Authorized to Access the HCP Data

There are a few steps you will need to take to get authorization to access the HCP data hosted on AWS.

You need to:

Complete the HCP Registration and agree to the HCP Open Access Data Use Terms

Complete the HBM Hackathon Registration

Create an Amazon Web Services Account

Email hbm.hackathon@gmail.com with your HCP and AWS email address(es)

A foreword from David van Essen:

For distribution to all investigators interested in using HCP Connectome-in-a-Box data.

IMPORTANT NOTICE to investigators wanting to use HCP datasets available on Connectome-in-a-Box hard drives.

HCP’s Connectome-in-a-Box provides imaging data from the Open Access dataset. Before using any of these data for research, you and all other investigators using the data are required to register and agree to the Open Access Data Use Terms. This includes agreeing to comply with institutional rules and regulations. This may mean that you need your research to be approved or declared exempt by a committee that oversees research on human subjects (e.g., your IRB or Ethics Committee). The released HCP data are not considered de-identified, insofar as certain combinations of HCP Restricted Data (available through a separate process) might allow identification of individuals. Different committees operate under different national, state and local laws and may interpret regulations differently, so it is important to ask about this. If needed and upon request, the HCP will provide a certificate stating that you have accepted the HCP Open Access Data Use Terms.

Sincerely,

David C. Van Essen (PI), for the WU-Minn HCP Consortium

May 1, 2013

Accessing HCP Data on AWS

Amazon Web Services is hosting the HCP Q1 Data Release as part of its Public Data Sets on AWS program, which will enable HBM Hackathon participants to get rapid access to the HCP data. Since the data is hosted on S3 in an uncompressed format, participants can download data in parallel using tools like s3cmd-modification.

You can use tools like s3cmd to list the contents of a directory on S3, get individual files or sync full directories. You will also be able to download directly to your personal/work computer or to a machine on the EC2 Cloud like the NITRC Computational Environment.

Getting Credentials

Ready to start downloading? Make sure you’ve completed the HCP registration and created an account with AWS.

Next, send an email to hbm.hackathon@gmail.com with your HCP and AWS email address(es).

We will use your HCP and AWS email address(es) to:

verify that you have agreed to the Open Access Data Use Agreement

grant you access to the HCP data on AWS.

Once verified, you will receive an email confirming that you have access to the HCP data on AWS.

Configuring your system

To access the data you’ll want to install s3cmd-modification, which will enable you to explore the HCP Data Directory on Amazon and will allow rapid parallel downloading (s3cmd-modification install instructions).

After you install s3cmd, you need to configure it with your AWS public and secret keys, located in AWS Security Credentials.

:~ s3cmd --configure

Enter new values or accept defaults in brackets with Enter.

Refer to user manual for detailed description of all options.



Access key and Secret key are your identifiers for Amazon S3

Access Key []: <your-access-key>

Secret Key []: <your-secret-key>

...

Take s3cmd out for a test drive…

List subject directories:

:~ s3cmd ls s3://hcp.aws.amazon.com/q1/



DIR   s3://hcp.aws.amazon.com/q1/100307/

DIR   s3://hcp.aws.amazon.com/q1/103515/

DIR   s3://hcp.aws.amazon.com/q1/111312/

...

DIR   s3://hcp.aws.amazon.com/q1/937160/

2013-04-17 06:57         0   s3://hcp.aws.amazon.com/q1/

Get a directory in parallel:

:~ s3cmd --parallel --workers=16 get --recursive s3://hcp.aws.amazon.com/q1/100307/T1w T1w



File s3://hcp.aws.amazon.com/q1/100307/T1w/BiasField_acpc_dc.nii.gz started [2 of 52]

File s3://hcp.aws.amazon.com/q1/100307/T1w/Native/100307.L.MyelinMap.native.func.gii started [3 of 52]

...

File s3://hcp.aws.amazon.com/q1/100307/T1w/T2w_acpc_dc_restore.nii.gz saved as 'T1w/T1w/T2w_acpc_dc_restore.nii.gz' (67855816 bytes in 143.3 seconds, 462.31 kB/s)

File s3://hcp.aws.amazon.com/q1/100307/T1w/BiasField_acpc_dc.nii.gz saved as 'T1w/T1w/BiasField_acpc_dc.nii.gz' (65076318 bytes in 180.3 seconds, 352.50 kB/s)

If everything checks out, you are ready to get hacking!