This title appears in the Scientific Report : 2016 
Challenges in Designing Workflows for Reproducible Analysis of Electrophysiological Data - Usage of Community Tools
Denker, Michael (Corresponding author)
Grün, Sonja
Theoretical Neuroscience; IAS-6
Computational and Systems Neuroscience; INM-6
2016
10.3389/conf.fninf.2016.20.00075
INCF Neuroinformatics 2016, Reading (United Kingdom), 2016-09-03 - 2016-09-04
Abstract
Kausative Mechanismen mesoskopischer Aktivitätsmuster in der auditorischen Kategorien-Diskrimination
Optogenetische Analyse der für kognitive Fähigkeiten zuständigen präfrontal-hippokampalen Netzwerke in der Entwicklung
Supercomputing and Modelling for the Human Brain
The Human Brain Project
Theory, modelling and simulation
Please use the identifier: http://dx.doi.org/10.3389/conf.fninf.2016.20.00075 in citations.


The complexity of workflows that span from the experimental recording of neuronal data up to the publication of final results of the analysis is increasing as datasets grow in size and analysis methods for massively parallel data sets become more intricate. The implementations of such internally designed workflows in various laboratories are highly heterogenous and far from being automatized. Software supporting such workflows is still largely under development or even missing. This situation leads to serious consequences regarding the degree of reproducibility of data capture and data analysis. Although the problem is well-known and leads to ineffective and unsustainable science, the development of guidelines and tools to facilitate a complete, provenance-tracked, and yet user-friendly workflow is an ongoing matter of research. Here, we sketch the primary challenges that complicate the design of workflows for electrophysiological research as we have witnessed them in our own work through collaborations with experimental groups. We outline how existing software tools can be combined and integrated to form partial workflows, and how such an approach is able to address some of the challenges (outlined in [1]). In particular, we practically demonstrate how a specific set of tools can be linked to a workflow with a standardized, yet flexible foundation, that builds on the specificities of the dataset, the analysis approach, and the available infrastructure (e.g., the available computational resources). We therefore identified and partly developed for this purpose three open-source tools that build the scaffold for a basic workflow: metadata management (odML, [2,3]), data representation (Neo, [4]), and data analysis (Elephant [5]). On the basis of this concrete workflow implementation we discuss open questions and urgently needed software components to tackle the challenge of reproducibility in the analysis of electrophysiological data. In part, this work is described in [6].AcknowledgementsThis work was supported by the Helmholtz Portfolio Theme Supercomputing and Modeling for the Human Brain (SMHB), EU grant 604102 (Human Brain Project, HBP) and Priority Program 1665 of the DFG (DE 2175/1-1 and GR 1753/4-1).References[1] Badia, R., Davison, A., Denker, M., Giesler, A., Gosh, S., Goble, C., Grewe, J., Grün, S., Hatsopoulos, N., LeFranc, Y., Muller, J., Pröpper, R., Teeters, J., Wachtler, T., Weeks, M. & Zehl, L. (2015) INCF Program on Standards for data sharing: new perspectives on workflows and data management for the analysis of electrophysiological data. Techn. Report, International Neuroinformatics Coordination Facility (INCF). https://www.incf.org/about-us/history/incf-scientific-workshops[2] Grewe, J., Wachtler, T. & Benda, J. (2011) A Bottom-up Approach to Data Annotation in Neurophysiology. Frontiers in Neuroinformatics 5, 16.[3] Zehl, L., Jaillet, F., Stoewer, A., Grewe, J., Sobolev, A., Wachtler, T., Brochier, T., Riehle, A., Denker, M. & Grün, S. Handling Metadata in a Neurophysiology Laboratory. Frontiers in Neuroinformatics (under revision).[4] Garcia, S., Guarino, D., Jaillet, F., Jennings, T., Pröpper, R., Rautenberg, P.L., Rodgers, C.C., Sobolev, A., Wachtler, T., Yger, P., et al. (2014) Neo: an object model for handling electrophysiology data in multiple formats. Frontiers in Neuroinformatics 8, 10.[5] Denker, M., Yegenoglu, A., Holstein, D., Torre, E., Jennings, T., Davison, A. & Grün, S. (2015) Elephant: An open-source tool for the analysis of electrophysiological data. Proceedings of the 11th Meeting of the German Neuroscience Society, Neuroforum 2015, T27-2B.[6] Denker, M. & Grün, S. Designing workflows for the reproducible Analysis of Electrophysiological Data. In: Brain Inspired Computing, eds: Katrin Amunts, Lucio Grandinetti, Thomas Lippert, Nicolai Petkov. Lecture Notes in Computer Science, Springer (in press).