Add electrical series description and resampling

[jthermiz]

Added resampling software (using process_nwb) so that user can resample down to nearest kHz if they desire. By default, it will resample down to nearest kHz if the hardware sample rate is not divisible by 1000.

I also added more description per Ryan's feedback.

One issue question I had is whether there's a better way to copy or modify the electrical series? Currently, I create a new one and copy over the electrical series attributes that I believe are important / used. But I'm not sure if I captured all of them.

I implemented these changes in neural_data_originator.py so that these operations can be used for any neural manager (TDT, HTK, etc...). @JesseLivezey suggested that the electrical series creation should happen in neural_data_originator and not within the neural managers.

[jihyunbak]

@jthermiz Thanks for the PR!

I haven't reviewed the resampling part yet, but here's my suggestion as to where we set the description text. I suggest we return to the previous structure in which the e-series is created at the Manager level, but

• create the description string in NeuralDataOriginator,
• and pass it as an additional keyword argument to the Manager.extract() method, like self.neural_data_manager.extract(device_name, dev_conf, description=description).

I think this also fits Ryan's suggestion in the Slack channel (pass a dictionary of attributes), because if we want to pass more attributes, the keyword arguments can be easily collected into a dictionary. How does this sound?

[jihyunbak]

Now about resampling … if the e-series is created at the Manager level, looks like resampling should also happen at that level, to avoid creating e-series twice. Perhaps a clean solution is to make a new class NeuralDataManager that has an empty extract() method, and a resample_es() method that does what you implemented here. Then HTKManager and TDTManager can inherit NeuralDataManager and implement their own versions of extract() as they do now.

~Finally, in NeuralDataOriginator, we can call self.neural_data_manager.resample_es() when the resampling flag is true.~ EDIT: we should probably pass the resampling flag as an kwarg to NeuralDataManager.

How does this sound? @jthermiz @JesseLivezey

[jihyunbak]

(based on the conversation on the Slack channel, I'm revoking most of my two comments above)

[jthermiz]

This PR has be superseded by #78