Saving minian output as a matlab file

[ph.va...@gmail.com]

Hi all,

I used the minian pipeline to analyse one of my sessions and I am really happy with the results (based on the final viaualisation), so first of all - big thank you to the developer(s) for sharing this tool with the community!

This is, however, more or less the first time I use python and I am a bit lost on how to convert the output to a Matlab file. What I want to get is three matlab matrices containing the A, C and S data from the pipeline.

I found this in the docs: https://minian.readthedocs.io/en/stable/tips/variables.html#saving-to-other-formats

However, when I try to run:

minian_ds = open_minian(dpath)

I get:

GroupNotFoundError: group not found at path ''

I tried:

minian_ds = open_minian(dpath + '/minian')

And I got a minian_ds var of type xarray.core.dataset.Dataset

However, I am a bit lost as to how to proceed. The final line in the docs says run

minian_ds["C"].values.save("C.npy") to get the npy and then convert to matlab but it gives me 

AttributeError: 'numpy.ndarray' object has no attribute 'save'

I hope it's as simple as I think it is and someone could lend a hand!

Best,

PV

 

[Federico Sangiuliano]

Hello PV. One way of loading the output from Minian in MATLAB is to save the output in the netCDF format. To achieve this, at the bottom of the Jupyter notebook you have to add a cell with the following code

minian_ds = open_minian(minian_ds_path)

minian_ds.to_netcdf("minian_dataset.nc")

This should save you a file named "minian_dataset" on the same folder where the output videos are located. Then, in MATLAB you can load the data like this

cd YOUR_PATH_TO_minian_dataset.nc

       
Adata  = ncread('minian_dataset.nc','A');                 
Cdata  = ncread('minian_dataset.nc','C');

Sdata  = ncread('minian_dataset.nc','S');

As an example, I processed three videos from my own recordings with the v4 Miniscope. Each video has 1000 frames, so in total I processed 3000 frames with a size of 608x608 pixels. It's easy to check this information from the notebook output of "varr"

By processing my data I got 606 neurons. Adata will hold the footprint of each neuron. Running "whos Adata" will give the following information

Name         Size                      Bytes  Class     Attributes

  Adata      608x608x606            1792131072  double

The structure of Adata is Adata(x,y,cell_number) where x and y will go from 1 to 608 and cell_number from 1 to 606. For C, the output from "whos Cdata" results in

Name          Size                Bytes  Class     Attributes

  Cdata      3000x606            14544000  double

The structure of Cdata is Cdata(amplitude_frame,cell_number), where amplitude_frame will be the amplitude of C in that frame for a given cell_number.

I hope this helps!

Federico

[Philip Vassilev]

Thank you, Federico, worked like a charm, it's exactly what I was looking for!

Kind regards,

PV

--
You received this message because you are subscribed to a topic in the Google Groups "Miniscope" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/miniscope/fW7xGqWLd4E/unsubscribe.
To unsubscribe from this group and all its topics, send an email to miniscope+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/miniscope/bcf68dff-b93f-47b8-8755-45eeabf32adfn%40googlegroups.com.

[Mark Sanders]

function MinianNetCDFtoMatlab()
%Grazie mille Federico Sangiuliano!


%% SELECT FOLDER WITH DATA
%folder = ['C:\Users\huberlab\Desktop\attempt 3'];
NCDFfolder = 'C:\Users\huberlab\Desktop\Bruno rig tests\Vibration_test_bruno_rig_1\ZB_81\customEntValHere\2022_03_10\20_59_58\My_V4_Miniscope\analsis attempts\4';

%look at what is contained in your data!
ncdisp([NCDFfolder '\' 'minian_dataset.nc']);

%The most usefull data will be saved in the following code
%see below which variables are not extracted from the .nc file

%% Read in tha data as Matlab variables
% files with xdata include all units, even those excluded in
% minian.visualization.CNMFViewer(), see label

Adata  = ncread([NCDFfolder '\' 'minian_dataset.nc'],'A');                  
Cdata  = ncread([NCDFfolder '\' 'minian_dataset.nc'],'C');
Sdata  = ncread([NCDFfolder '\' 'minian_dataset.nc'],'S');
c0data = ncread([NCDFfolder '\' 'minian_dataset.nc'],'c0');
b = ncread([NCDFfolder '\' 'minian_dataset.nc'],'b');
b0 = ncread([NCDFfolder '\' 'minian_dataset.nc'],'b0');
f = ncread([NCDFfolder '\' 'minian_dataset.nc'],'f');
YrAdata = ncread([NCDFfolder '\' 'minian_dataset.nc'],'YrA');

units = ncread([NCDFfolder '\' 'minian_dataset.nc'],'unit_id');
unit_labels = ncread([NCDFfolder '\' 'minian_dataset.nc'],'unit_labels');

max_proj = ncread([NCDFfolder '\' 'minian_dataset.nc'],'max_proj');


%Less usefull variables: height, width, frame, motion
% height = ncread([NCDFfolder '\' 'minian_dataset.nc'],'height');
% width = ncread([NCDFfolder '\' 'minian_dataset.nc'],'width');
% motion = ncread([NCDFfolder '\' 'minian_dataset.nc'],'motion');

%Variables; animal, session, shift_dim are of an unsupported datatype and
% cannot be extracted with ncread


%% Make version of variables with bad cells (label =-1) removed
indx = 0 <= unit_labels;

ids = unit_labels(indx);
A = Adata(:,:,indx);
C = Cdata(:,indx);
S = Sdata(:,indx);
c0 = c0data(:,indx);
YrA = YrAdata(:,indx);

clear indx

%% Saves it to a .mat file in the same folder
save([NCDFfolder '\' 'netCDFData.mat'])

end

Op zaterdag 8 januari 2022 om 11:09:14 UTC+1 schreef ph.va...@gmail.com: