research data management for individual scientists Ivan Zimine 2013-06-14 DACETS

$ whoami Physicist ? Neuroscientist ? Programmer ? Manager

https://speakerdeck.com/kennethreitz/api-driven-development

Open Science & Reproducible Research open {access, source, data}

CERN, LHC

The mission of the WLCG project is to provide global computing resources to store, distribute and analyse the ~25 Petabytes (25 million Gigabytes) of data annually generated by the Large Hadron Collider (LHC) at CERN on the Franco-Swiss border.

60 hrs of video / min 60 *100MB * 60min * 24 = 8TB/day (3 PB/year)

BIG TOYS SMALL TOYS

BIG TOYS SMALL TOYS

Research Data publications raw data derived data summary data annotations analysis workflow code/scripts...

... brain variability (an old project...)

original brain extract brain split GM/WM segm sulci skeleton simple surfaces 3D recon & labeling data processing pipeline

data organization ivan@pavlov:~/mri_data/fMRI_Lang_var $ tree -L 2 . ├── Anat_variability/ │ ├── 01_GRV/ │ ├── 02_GLZ/ │ ├── 03_LIB/ │ ├── 04_RAC/ │ ├── 05_DUB/ │ └── 06_WIL/ └── fMRI_Lang/ ├── 01_GRV/ ├── 02_GLZ/ ├── 03_LIB/ ├── 04_RAC/ ├── 05_DUB/ └── 06_WIL/ ivan@pavlov:~/mri_data/fMRI_Lang_var $ tree fMRI_Lang/ fMRI_Lang/ ├── 01_GRV/ │ ├── analyze/ │ │ ├── fmri_noun.hdr │ │ ├── fmri_noun.img │ │ ├── fmri_verb.hdr │ │ ├── fmri_verb.img │ │ ├── t1_3d.hdr │ │ ├── t1_3d.img │ │ ├── t1_axe.hdr │ │ └── t1_axe.img │ ├── maps/ │ │ ├── zscore_noun.hdr │ │ ├── zscore_noun.img │ │ ├── zscore_verb.hdr │ │ └── zscore_verb.img │ └── raw/ │ ├── DICOM/ │ │ ├── 0000001.ima │ │ ├── 0000002.ima │ │ └── ... (~1000 files) │ └── DICOMDIR ├── 02_GLZ/ ├── 03_LIB/ original project image sequence header file derived project

data organization ivan@pavlov:~/mri_data/fMRI_Lang_var $ tree fMRI_Lang/ fMRI_Lang/ ├── 01_GRV/ │ ├── analyze/ │ │ ├── fmri_noun.hdr │ │ ├── fmri_noun.img │ │ ├── fmri_verb.hdr │ │ ├── fmri_verb.img │ │ ├── t1_3d.hdr │ │ ├── t1_3d.img │ │ ├── t1_axe.hdr │ │ └── t1_axe.img │ ├── maps/ │ │ ├── zscore_noun.hdr │ │ ├── zscore_noun.img │ │ ├── zscore_verb.hdr │ │ └── zscore_verb.img │ └── raw/ │ ├── DICOM/ │ │ ├── 0000001.ima │ │ ├── 0000002.ima │ │ └── ... (~1000 files) │ └── DICOMDIR ├── 02_GLZ/ ├── 03_LIB/ ivan@pavlov:~/mri_data/fMRI_Lang_var $ tree Anat_variability/ Anat_variability/ ├── 01_GRV/ │ ├── READMEs/ │ ├── anatomy/ │ │ ├── T1.hdr │ │ ├── T1.img │ │ ├── T1_norm.hdr │ │ └── T1_norm.img │ ├── graphe/ │ ├── segment/ │ │ ├── T1_brain.hdr │ │ └── T1_brain.img │ └── tri/ │ ├── LHemi.dat │ └── RHemi.dat ├── 02_GLZ/ ├── 03_LIB/ ├── 04_RAC/ ├── 05_DUB/ └── 06_WIL/

data organization $ find . -name \*3d.img -exec script.py {}\; useful, but clearly not enough (binary headers)

what’s the problem? I can not reproduce my own research results

what’s the actual problem? lack of simple tools for handling collections of [large] binary objects

what’s needed? help with data org structure tracking notes/comments about data tracking provenance search (beyond filename/{c,m}time) efficient local/remote sync ...

isn’t it a solved problem? http://wlcg.web.cern.ch/ http://www.mygrid.org.uk/ http://galaxyproject.org/ https://www.globusonline.org/ https://www.opensciencegrid.org/ http://www.openmicroscopy.org/

No content

“Before someone suggests OME, we don’t have the wherewithal to move to OMERO – the server setup is beyond me and not something that we can implement easily through our IT support. This is one for the future…”

No content

No content

isn’t it a solved problem? Amazon AMIs Media management (iTunes et al.) SW packages (deb, rpm, egg, npm,...)

setup.py setup(name='beets', version='1.1.1', description='music tagger and library organizer', author='Adrian Sampson', author_email='[email protected]', url='http://beets.radbox.org/', license='MIT', long_description=_read('README.rst'), packages=[ 'beets', 'beets.ui', ] install_requires=[ 'mutagen>=1.21', 'musicbrainzngs>=0.4', 'pyyaml', ] classifiers=[ 'Topic :: Multimedia :: Sound/Audio', 'License :: OSI Approved :: MIT License', 'Environment :: Console', 'Programming Language :: Python :: 2.7', ], )

package.json { "name": "http-server", "version": "0.3.0", "description": "a simple command-line http server", "license": "MIT", "author": "Nodejitsu ", "contributors": [ { "name": "Marak Squires", "email": "[email protected]" } ], "repository": { "type": "git", "url": "https://github.com/nodejitsu/http-server.git" }, "keywords": [ "cli", "http", "server" ], "dependencies" : { "flatiron" : "0.1.x", } }

No content

data-package.json { # general "metadata" name: "a-unique-human-readable-and-url-usable-identifier", title: "A nice title", licenses: [...], sources: [...], resources: [ { ... resource info described below ... } ], # optional "metadata" ... additional information ... }

data-package.yaml --- project: name: "project name" license: ... description: ... resources: ... experiment: sample: "sample_id", date: "2013-01-01 ", equipment: ... acquisition: # (simulation, data-analysis) settings: ... data: - path: url

dacets $ dct import /path/to/project/ $ dct add /path/to/file $ dct remove /path/to/file $ dct find “mri+brain+3T+T1” -o dacets.json $ dct sync -f dacet.json >>> from dacets import Dacets >>> dc1 = Dacets.load(‘dacets.json’) >>> dc2 = Dacets.find(‘mri+brain+3T+T1’) >>> run_pipeline(on=dc2)

dacets datasets + facets binary files DVCS python module (integration with ipython notebook)

interesting tech git-annex hg LargefilesExtension http://neuralensemble.org/sumatra/ https://github.com/okfn/dpm/