link to homepage

Institute of Neuroscience and Medicine

Navigation and service

Team Data Science of Electro- and Optophysiology in Behavioural Neuroscience (DSEO)

Team Leader: Dr. Michael Denker

Computational neuroscience is focused on gaining an understanding of the principles of information processing in the brain. This involves the development of theories and models of brain function, as well as the design of analysis methods to probe for signatures of the corresponding dynamics in data from the brain. Models and experimental data cover various spatio-temporal scales, are represented by multiple modalities of observation, and address a diversity of neuroscientific concepts expressed on varying levels of abstraction. At the same time, neuroscientists have a diversified and constantly growing repertoire of methodologies to analyse their data, at the cost of higher complexity and computational expense. The team Data Science for Electro- and Optophysiology Behavioural Neuroscience (DSEO) of Dr. Michael Denker at the JARA Institute for Brain-Function Relationships (INM-10) and Institute for Computational and Systems Neuroscience (INM-6) addresses the challenges that arise from this situation in the context of data management and data processing by combining knowledge of the neuroscience disciplines with concepts from physics and computer science.

Shareable, standardised, in-depth descriptions of activity data sets and their origin

In the past decades, neuroscientists have witnessed a rapid increase in the complexity and volume of data, including, in particular, the subdomains of electrophysiology and optophysiology. Today, electro- and optophysiological experiments combine recordings of hundreds of channels in parallel with behaviour under increasingly natural conditions. Minute details of the experiment may become relevant for the analysis of such rich data, often in ways that are hardly foreseen during the time when the experiment is conducted. Despite the importance of unambiguous and machine-readable metadata describing the experiments, current workflows for acquisition and post-processing are largely homegrown, use custom data and metadata representations, are difficult to re-use, and rely on error-prone manual intervention. The team addresses the need for defining practical solutions to automate, standardise, and streamline data and metadata acquisition in the experimental laboratory (Zehl et al., 2016). This process leads to the identification and implementation of missing software components (Sprenger et al., 2019) for digitised acquisition workflows that assist scientists in creating high-quality publications of their precious data (Brochier et al, 2018). To expedite bringing concepts of FAIR data management to wide-spread use in the laboratories, the team is among the initial members to spark the NFDI Neuroscience consortium ( as a community for exchange of knowledge on research data management.

  • Brochier, T., Zehl, L., Hao, Y., Duret, M., Sprenger, J., Denker, M., Grün, S., Riehle, A., 2018. Massively Parallel Recordings in Macaque Motor Cortex during an Instructed Delayed Reach-to-Grasp Task. Scientific Data 5, 180055.
  • Sprenger, J., Zehl, L., Pick, J., Sonntag, M., Grewe, J., Wachtler, T., Grün, S., Denker, M., 2019. odMLtables: A User-Friendly Approach for Managing Metadata of Neurophysiological Experiments. Front. Neuroinform. 13, 62.
  • Zehl, L., Jaillet, F., Stoewer, A., Grewe, J., Sobolev, A., Wachtler, T., Brochier, T.G., Riehle, A., Denker, M., Grün, S., 2016. Handling Metadata in a Neurophysiology Laboratory. Frontiers in Neuroinformatics 10, 26.

Designing workflows for reproducible data acquisition and post-processing. Figure modified from: Zehl et al. (2016).Designing workflows for reproducible data acquisition and post-processing. Figure modified from: Zehl et al. (2016).

Standardisation of complex analysis processes across models and experiments

The lack of standardisation in data acquisition also affects the downstream analysis process. With the development of increasingly sophisticated methodological approaches, a well-defined, traceable, and reproducible analysis pipeline is essential to strengthen confidence of analysis results, facilitate joint collaborative analysis on the same dataset, and enable the smooth transition from early interactive exploratory analysis processes to automated processing of large data volumes using high performance computing resources (Bouchard et al., 2016; Denker and Grün, 2016). Collaboratively, the team drives forward efforts coordinated through EBRAINS ( to develop open-source community-centred elements for creating such analysis workflows, such as the Electrophysiology Analysis Toolkit (Elephant) for the analysis of concerted dynamics exhibited by spike train data and population signals ( and Neo for unified data representations in simulations and experiments (

  • Bouchard, K.E., Aimone, J.B., Chun, M., Dean, T., Denker, M., Diesmann, M., Donofrio, D.D., Frank, L.M., Kasthuri, N., Koch, C., Ruebel, O., Simon, H.D., Sommer, F.T., Prabhat, 2016. High-Performance Computing in Neuroscience for Data-Driven Discovery, Integration, and Dissemination. Neuron 92, 628–631.
  • Denker, M., Grün, S., 2016. Designing Workflows for the Reproducible Analysis of Electrophysiological Data, in: Amunts, K., Grandinetti, L., Lippert, T., Petkov, N. (Eds.), Brain-Inspired Computing. Springer International Publishing, Cham, pp. 58–72.

Opens new windowPerforming analysis of neuronal dynamics using the Elephant and Neo Python libraries (cf.,

Advancing processes for verification and validation in neuroscience

As the complexity of research topics in neuroscience grows, so does the urgency for rigorous methods of verification (does a computational process produce the correct result?) and validation (does a model represent features of the system of interest?). In the context of tools for data analysis, the team considers reproducibility of prior scientific findings as a major element of verification and validation processes (Rostami et al., 2017). With regard to model validation, the team is engaged with conceptual work as well as concrete implementations of software (NetworkUnit, to measure the degree of agreement between neuronal network simulations and experimental data (Trensch et al., 2018; Gutzen et al., 2018). To build valid validation processes that exploit the richness of data types obtained across experimental techniques and that make use of the diversity of analysis approaches, the team investigates the design of multi-modal, multi-methodological, interoperable workflows (see video here).

  • Gutzen, R., von Papen, M., Trensch, G., Quaglio, P., Grün, S., Denker, M., 2018. Reproducible Neural Network Simulations: Statistical Methods for Model Validation on the Level of Network Activity Data. Frontiers in Neuroinformatics 12, 90.
  • Rostami, V., Ito, J., Denker, M., Grün, S., 2017. [Re] Spike Synchronization And Rate Modulation Differentially Involved In Motor Cortical Function. ReScience 3, 3.
  • Trensch, G., Gutzen, R., Blundell, I., Denker, M., Morrison, A., 2018. Rigorous Neural Network Simulations: A Model Substantiation Methodology for Increasing the Correctness of Simulation Results in the Absence of Experimental Validation Data. Frontiers in Neuroinformatics 12, 81.

Formalisation of model verification and validation processes (left) as implemented in the framework of the SciUnit-based NetworkUnit validation library. Figures from Gutzen et al. (2018).Formalisation of model verification and validation processes (left) as implemented in the framework of the SciUnit-based NetworkUnit validation library. Figures from Gutzen et al. (2018).