Help understanding `get_data_in_units` in `pynwb` for `ElectricalSeries` #83

rohanshah · 2024-06-21T15:01:45Z

rohanshah
Jun 21, 2024

Hello, I hope this is the appropriate for getting help understanding some source code in pynwb. Happy to file an issue in that GitHub repository instead if that is more appropriate.

I was reading through some of the source code for the TimeSeries and ElectricalSeries object and was confused by the implementation of get_data_in_units... hoping to clarify my understanding with this discussion post.

It seems that when channel_conversion is present, the get_data_in_units function tries to multiply the conversion factor across the time dimension instead of the electrode dimension?

For example, if I have data with a shape of (100, 2) (i.e. 100 samples across two electrodes) and a channel_conversion = channel_conversion = np.asarray([2.0, 3.0]) (i.e. a scaling factor of 2 and 3 for each electrode, respectively), I get the error

ValueError: operands could not be broadcast together with shapes (100,2) (2,1)

If I change the source code to perform this calculation then it seems to correctly scale the data (e.g. all samples for the first electrode multiplied by a factor of 2 and all samples for the second electrode multiplied by a factor of 3)

            scale_factor = self.conversion * self.channel_conversion

Further adding to the confusion is the unit test for this case:

https://github.com/NeurodataWithoutBorders/pynwb/blob/90e60484415d38d3d566dc9fea63b6d99ffdbe43/tests/unit/test_ecephys.py#L119-L135

Specifically, the first few lines of setup:

        data = np.asarray([[1, 1, 1, 1, 1], [1, 1, 1, 1, 1]])
        conversion = 1.0
        offset = 3.0
        channel_conversion = np.asarray([2.0, 2.0])

It was my understanding of the shape of data (i.e. (2, 5)) that it is meant to represent 5 electrodes with 2 samples each but the channel_conversion is for 2 channels...

Maybe I am misunderstanding the way data or channel_conversion are meant to be shaped?

Thank you for your help!