Example datasets

This module provides a simple interface to load data used in some Elephant tutorials and that can also be used as example neuronal activity data for learning, testing and development purposes.

Each dataset can be loaded by the load_data() function by providing a specific dataset name. Some are downloaded from the official elephant-data G-Node GIN repository and others are generated by simulation. For downloaded files, the function also takes care of any loading needed, directly returning the relevant data objects. For generated data, the required parameterization is already defined.

Currently, the following data are available for loading:

  • asset: a dataset containing the activity of 500 parallel spike trains from a simulation of synfire chains that are activated repeatedly (Schrader et al., 2008). The data is loaded as a single neo.core.Segment object, containing the 500 neo.core.SpikeTrain objects with the spiking activity of each neuron (accessible by the .spiketrains attribute). This dataset is used in the tutorial for the Analysis of Sequences of Synchronous EvenTs (ASSET) method (Torre et al., 2016).

  • granger_causality_indirect: a dataset of three simulated time series X, Y, and Z, with indirect causal influence from Y to X through Z (Y -> Z -> X). Each time series has 10000 sample points. The data is loaded as a (10000, 3) numpy.ndarray object. The second dimension is the time series dimension, where each column corresponds to one of the three time series, ordered as X, Y, Z. This data is used in the tutorial for the conditional Granger causality method and it is recreated from Example 2 section 5.2 of Ding et al. (2006).

  • granger_causality_both: a dataset of three simulated time series X, Y, and Z, with both direct and indirect causal influences from Y to X (Y -> X and Y -> Z -> X). Each time series has 10000 sample points. The data is loaded as a (10000, 3) numpy.ndarray object. The second dimension is the time series dimension, where each column corresponds to one of the three time series, ordered as X, Y, Z. This data is used in the tutorial for the conditional Granger causality method and it is recreated from Example 2 section 5.2 of Ding et al. (2006).

  • unitary_events: a dataset containing the simultaneously recorded activities of two neurons in the primary motor cortex of monkeys performing a delayed-pointing task (Riehle et al., 1997). The neuronal activity is recorded across 36 trials. The dataset consists of a list with 36 inner lists (one per trial). Each trial list contains two neo.core.SpikeTrain objects storing the spike times of each neuron. This dataset is used in the tutorial for the Unitary Event Analysis method (Grün et al., 1999), which detects the coordinated spiking activity that occurs significantly more often than predicted by the firing rates of the neurons.

load_data(name)

This function loads example data used in Elephant tutorials and examples.

References

[Schrader, 2008]

S. Schrader, S. Grün, M. Diesmann, and G. L. Gerstein. Detecting synfire chain activity using massively parallel spike train recording. Journal of neurophysiology, 100(4):2165–2176, 2008. doi:10.1152/jn.01245.2007.

[Torre, 2016]

E. Torre, C. Canova, M. Denker, G. Gerstein, M. Helias, and S. Grün. Asset: analysis of sequences of synchronous events in massively parallel spike trains. PLoS Comp. Biol., 12(7):e1004939, 2016. doi:10.1371/journal.pcbi.1004939.

[Ding, 2006] (1,2)

Mingzhou Ding, Yonghong Chen, and Steven L. Bressler. Granger causality: basic theory and application to neuroscience. 2006. arXiv:q-bio/0608035.

[Riehle, 1997]

A. Riehle, S. Grün, M. Diesmann, and A. Aertsen. Spike synchronization and rate modulation differentially involved in motor cortical function. Science, 278(5345):1950–1953, 1997. doi:10.1126/science.278.5345.1950.

[Gr{\"u}n, 1999]

S. Grün, M. Diesmann, F. Grammont, A. Riehle, and A. Aertsen. Detecting unitary events without discretization of time. J. Neurosci. Meth., 94(1):67–79, 1999. doi:10.1016/s0165-0270(99)00126-0.