Neural Data Science for Vision Research
Technology and Development
Spike inference toolbox
Two-photon calcium imaging has become a major tool to study neural population activity. Neural firing rates, however, can only be indirectly inferred from the measured traces. We have developed an algorithm based on supervised training of flexible probabilistic models and have shown that it outperforms other published techniques (link). The toolbox is available in Python, including a pre-trained model.
Circular statistics toolbox
Angular data requires special statistical methods. We have developed a toolbox for circular statistics containing specialized methods for a large range of statistical problems involving angular data. The toolbox is available in Matlab and Python. There also is a companion paper describing the toolbox
Large and complex datasets in neuroscience require special attention to data management. DataJoint implements an easily adaptable front-end to a relation database system for flexible and reliable data analysis. In particular, the system ensures data integrity and supports parallel computation. It is available in Matlab and Python.
Mouse RGC dataset
We recorded responses from >11,000 retinal ganglion cells in the mouse retina to a standardized stimulus set using two photon imaging. The data was used to provide a functional clustering of the different RGC types in the mouse. For a detailed description of the experimental procedure, see Baden et al.
Macaque V1 dataset
We recorded responses of several hundreds of neurons in V1 of awake macaques to static and moving gratings. The neurons were recorded using tetrode arrays, while the monkey was required to maintain fixation. The dataset contains parallel recordings of 5-20 neurons.