pyPhotometry Figure3f

[Yifang Yuan]

Hi,

I was trying to replicate the testing on the pyPhotometry paper, Figure 3f, with time-division mode, but I can hardly get similar coefficient of variation as the original paper. Mine are 5-10 times larger than the paper. 

Has anyone done the same test as FIgure3f with your board? 

If  @Akam could see this, would you let me know the STD values in the Figure 3f?

Here is our test, we used DiI, but maybe the DiI solution is not enough so our signal is small by itself (0.6-1.2V on GUI)?  So the mean signal value in our case is much smaller, so I'd like to ask your mean and STD values...

I didn't compare TD with FD but with the continous mode. 

The x-axis is the absolute value of the current when the LED is on (the current on the GUI). I think in the original paper, you use the averaged LED current value, so our x-axis should be comparable with the paper. 

Note that we have 120mA data because we changed a resistor to get larger current.

[thoma...@neuro.fchampalimaud.org]

Hi Yifang,

The full data and analysis code used to generate the paper figures are at https://github.com/pyPhotometry/manuscript so the best way to work out what the discrepancy is between your measurements and mine is probably to download my data and code so you can look in detail at each step of the analysis.  The analysis for figure 3f is implemented by the function FD_TD_noise_comparison at line 369 in figures.py.

One possible reason for the discrepancy would be if you are lowpass filtering your signals differently.  For the paper figure 3F the signals were low pass filtered at 20Hz using a 2nd order butterworth filter run over the data forward and backward using scipy's filtfilt function. 

Another possibility is that you have noise artifacts in your signals.  Photorecievers can be sensitive to electrical noise and we have found that mobile phones can generate large artifacts in the signals if they are close to the photoreciever.   This would show up as big spikes when you plot the raw signals.

Let me know what you find,

Thomas

[Yifang Yuan]

Hi Thoma,

Thank you so much for the fast response. 

I refered to your code so the lowpass filter should be the same. I just checked your original data, I think the magnitudes of signal means are quite different, with the same LED current, we get ~0.2V in our case while 2V in yours. So even if the STDs are similar, the CoV(STD/mean) will be very different. We'll get some DiI to test again. 

I agree with the electrical noise, I used to find a 7Hz noise in both channels and don't know where it comes from...so we're trying to isolate the system with other devices.

Best,

Yifang

[thoma...@neuro.fchampalimaud.org]

That makes sense, glad you were able to work out the discrepancy.