How to Convert .CSV and .MAT file into .NWB file?

[Anannabiswas]

Hello, I am quite new in Dandi community. I don't have much knowledge regarding the nwb file. I want to upload my data set (.csv and .mat files) into the dandi. As of my understanding I need to convert it into nwb file for uploading and publishing it into Dandi archive.

I wanted to know how to convert my .csv and .mat file into NWB files. Moreover, the csv and mat files are generated from the computational model. Would you kindly help me with the conversion?

My current dataset that I need to upload into dandi. The link is here: https://github.com/Brain-Inspired-AI-Lab/Parkinson-Electrophysiological-Signal-Dataset-PESD

Thanks a lot in advance!

[kabilar]

Hi @h-mayorquin, could you please provide an explanation to @Anannabiswas? Thanks.

[h-mayorquin]

Hi @Anannabiswas, am I understanding correctly that your data comes from simulations?

Taking a look at your data I only see the controller_beta_values.csv and the matlab file. As we don't have access to the matlab files and the controller_beta_values.csv are not self-explanatory, would you mind providing a description of what type of data they are and what do they stand for?

[Anannabiswas]

Yes, you are correct. The data is generated from a simulation.

Sure. The data (in matlab file: STN_LFP.mat) is generated from the cortico-basal ganglia computational model, represented as raw local field potentials (LFP). The raw LFP is collected from the electrical fluctuations generated by the synchronized activity of a population of neurons between the cortex, subthalamic nucleus (STN), and thalamus of the computational model. The unit of the data is (uV/sec).

For the (controller_beta_values.csv) file, I applied the bandpass filter to the raw LFP (STN_LFP.mat file) to get the beta band filtered LFP. Then I used the full-wave rectification of the beta band filtered LFP signal using a fourth-order Chebyshev band-pass filter with an 8 Hz bandwidth, centered about the peak in the LFP power spectrum. This is how I got the Beta Average rectified signal (ARV) which is represented here as controller_beta_values.csv. The unit of the data is (uV/sec).

More details about the computational model can be found in this paper, https://pubmed.ncbi.nlm.nih.gov/32194372/

Kindly let me know if you need more information or further clarification.

[Anannabiswas]

Would you kindly tell me how to convert the files (.csv and .mat) into nwb?
Thanks for your help.

[h-mayorquin]

Hi @Anannabiswas @kabilar,

Based on the description, I'm not sure this data complies with the policy. The archive is intended for empirical and not simulated data. More concretely, the policy states:

Status of research data: Empirical (not simulated) data and associated metadata from any stage of the research study's life cycle are accepted. Simulated data is handled on a case-by-case basis; contact the DANDI team.

Policy reference

That said, there is no standard procedure for converting either CSV or MATLAB files—it depends on the type of data you have. However, based on your description, I am fairly certain that this tutorial is what you need:

Neurodata Without Borders (NWB) Ecephys Tutorial

[kabilar]

Hi @Anannabiswas @h-mayorquin, I think you should be fine to post this simulated electrophysiology data to DANDI Archive, as I am not aware of another NIH BRAIN Initiative-sponsored archive where this data would be better suited. Just be sure to note in the description on the Dandiset landing page that this is simulated data. cc @satra

As @h-mayorquin mentioned, since your data is not from standard data acquisition or data analysis packages, you'll need to review the NWB tutorials and use either the PyNWB API or MatNWB API to create a NWB file with your data. Hope that helps.

[Anannabiswas]

Thank you so much for all the information. I'll review the NWB tutorials and format the data before submission using the PyNWB or MatNWB API. I appreciate all the guidance.