Datajongleur - controlling data objects and data models with Python at the neuroscientist's workbench
Filed under:
General neuroinformatics
Philipp Lothar Rautenberg (Ludwig-Maximilans Universität München, Department of Biology, G-Node, Planegg-Martinsried), Andrey Sobolev (Ludwig-Maximilans Universität München, Department of Biology, G-Node, Planegg-Martinsried), Michael Stransky (Ludwig-Maximilians-Universität München, Institute for Computer Science, Munich), Thomas Wachtler (Ludwig-Maximilans Universität München, Department of Biology, G-Node, Planegg-Martinsried)
Scientific progress depends increasingly on data management efforts that involve storing and structuring data, querying and analyzing data, exchange of data, and re-analysis of previously recorded data. This causes a major barrier to fully exploit the scientific potential of experimental data. In order to make data analysis, re-analysis, and data sharing efficient, data management has to start at the local workbench closely linked to the analysis methods used by the scientist.
We present the python package Datajongleur which supports the scientist in controlling single data objects when writing scripts and software with Python for analyzing scientific data. Datajongleur provides predefined objects representing scientific data, such as recorded signals together with corresponding units – but also objects representing domain-specific measurements like spike trains (domain of electrophysiology) or morphological skeleton trees (domain of morphology) . These objects can be loaded from, changed, and stored within a relational database like PostgreSQL [1] (server based) or SQLite [2] (file based). Furthermore, single data objects can be arranged within domain specific data models, thus adding information about the relations between data objects. Datajongleur handles domain specific data models as extensions. Some recently evolving data models are already implemented at a corresponding stage of development, such as neo [3] which is a data model with objects for the domain of electrophysiology, or libNeuroML [4] with object for the damain of morphology. In addition, scientists can write their own extensions if necessary.
Together, Datajongleur bridges a gap between persistent data storage realized with relational database technology on the one side and single data objects that can be arranged as data models on the other side. This facilitates data management for the scientist on the level of scripting and programming where data analysis takes place.
Supported by the Federal Ministry of Education and Research (BMBF, grant 01GQ0801 and grant 01GQ1116).
[1] http://www.postgresql.org/
[2] http://www.sqlite.org/
[3] http://neuralensemble.org/trac/neo
[4] https://github.com/NeuralEnsemble/libNeuroML/
We present the python package Datajongleur which supports the scientist in controlling single data objects when writing scripts and software with Python for analyzing scientific data. Datajongleur provides predefined objects representing scientific data, such as recorded signals together with corresponding units – but also objects representing domain-specific measurements like spike trains (domain of electrophysiology) or morphological skeleton trees (domain of morphology) . These objects can be loaded from, changed, and stored within a relational database like PostgreSQL [1] (server based) or SQLite [2] (file based). Furthermore, single data objects can be arranged within domain specific data models, thus adding information about the relations between data objects. Datajongleur handles domain specific data models as extensions. Some recently evolving data models are already implemented at a corresponding stage of development, such as neo [3] which is a data model with objects for the domain of electrophysiology, or libNeuroML [4] with object for the damain of morphology. In addition, scientists can write their own extensions if necessary.
Together, Datajongleur bridges a gap between persistent data storage realized with relational database technology on the one side and single data objects that can be arranged as data models on the other side. This facilitates data management for the scientist on the level of scripting and programming where data analysis takes place.
Supported by the Federal Ministry of Education and Research (BMBF, grant 01GQ0801 and grant 01GQ1116).
[1] http://www.postgresql.org/
[2] http://www.sqlite.org/
[3] http://neuralensemble.org/trac/neo
[4] https://github.com/NeuralEnsemble/libNeuroML/
Preferred presentation format:
Poster
Topic:
General neuroinformatics
Latest news for Neuroinformatics 2011
Follow INCF on Twitter
