Problems with array shape when running dPCA in Python

[franjmt]

Hi, I constructed the initial neural activity matrix (dF_dff) size (N, S, D, T, E): N = number of neurons S = number of stimuli D = number of decision T = time each trial E = max number of trial per condition

In my case dF_dff shape is (1307, 2, 2, 83, 43), and I take the average on the first dimension (neurons), obtaining dF_dff_average, shape (2, 2, 83, 43). To run the dPCA, I am doing the following thing:

label = 'tsd'
join = [{'s': ['s', 'ts']}, {'d': ['d', 'td']}, {'t'}, {'sd': ['tsd']}]
dpca = dPCA.dPCA(labels=label, join=join, n_components=2, regularizer='auto')
dpca.protect = ['t']
Z = dpca.fit_transform(dF_dff_average, dF_dff)

But this doesn't work. I get an index error:

IndexError: index 43 is out of bounds for axis 3 with size 43

Maybe I am doing something wrong. Hopefully you can help me figuring out what's the problem Thank you

[dkobak]

The average over neurons does not make any sense to me. Shouldn't it be the average over trials, i.e. the last dimension?

[franjmt]

Oh yes, I did the average over trials first, but still I am having the same issues. I think I got confuse with the python demo, because the trial is on axis 0, but the Matlab has it on axis 4. Regardless of that error, if I use:

dF_dff_average = np.nanmean(dF_dff, axis=4) dF_dff_average.shape (1307, 2, 2, 83) dF_dff.shape (1307, 2, 2, 83, 43)

[dkobak]

Well, if the Python code expects the trials to be on axis 0 then you should reorder your dimensions to make it such.

[franjmt]

Still if I move the trials to axis 0 and average over trial, still doesn't run.

[dkobak]

Please post the new shapes of your arrays and the error message you get.

[franjmt]

dF_dff.shape
(43, 1307, 2, 2, 83)

dF_dff_average.shape
(1307, 2, 2, 83)

label = 'tsd'
join = [{'s': ['s', 'ts']}, {'d': ['d', 'td']}, {'t'}, {'sd': ['tsd']}]

dpca = dPCA.dPCA(labels=label, join=join, n_components=2, regularizer='auto')
dpca.protect = ['t']

Z = dpca.fit_transform(dF_dff_average, dF_dff)

Error: ValueError: operands could not be broadcast together with shapes (1307,2,83,2) (1307,2,2,1)

[dkobak]

Okay, the neuron dimension should come 1st (after trials but before time/stim/decision). Please look in the demo!

[franjmt]

In the demo the time is in axis 2 of R data

# number of neurons, time-points and stimuli
N,T,S = 100,250,6

R.shape
(100, 6, 250)

[dkobak]

Right, yes, the dimensions seem fine now...

If time is last, label = 'tsd' should be 'sdt' probably. Not sure this can explain the error though.

[franjmt]

It worked. Thank you very much!

[dkobak]

Great!