Has anyone used the Simplified Up-Down method developed by Bonin et al.?
It seems to have many benefits, namely speed and simplicity.
Hi. I wanted to wait a bit before chiming in on this thread. We use SUDO in our lab (surprise), and I know a few other groups in Canada and the US that are using it. I have heard from one researcher that is reluctant to switch from Chaplan to SUDO because all their data were collected with Chaplan. From everything I have seen and analyzed, this worry is misplaced and the data from both methods have the same mean result, variance, etc. I’ve had new students get familiar with both methods and there are far more testing errors with Chaplan, just because there is more to keep track of. They also much preferred the speed and ease of SUDO, so now we just use SUDO exclusively.
I’m always happy to answer any questions about the method that people might have.
Hi @Boninrp. Thanks so much for joining the discussion!
I’ve been using SUDO now for a few weeks and overall I’m pleased. It’s easy and quick. In the end, whether it’s the Chaplan method or SUDO, any derivative of the Dixon method is an approximation of the 50% withdrawal threshold, and isn’t the actual ground truth. Doing a series of different force VF hairs and looking at the response rates, such as this paper (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4956152/), is probably going to get you a little closer to the truth, but then you greatly increase the labor and also, I’ve found that the mice do alter their behavior in response to the abundant pokes they get. They seem to get more sensitive over time.
Anyway, I’m using SUDO now. thanks for developing the method.
One thing… do you have a useful Excel sheet or anything that converts responses to final fiber number (or force in grams)?
@Boninrp
For publication, do you report the fiber number (1-9), or gram force (this one being much more common)?
I actually advocate the use of pressure (g/mm^2) rather than force (g) for PWT for a few reasons: the VF filament increases in diameter at high bending forces, so the bending force is applied over a larger area; VFs use a logarithmic scale of actual force to create a linear scale of perceived intensity; PWT in force is rarely normally distributed so “normal” statistics are not appropriate (t-test, ANOVA, etc) nor is the use of symmetrical SEM on graphs. A great, physiological argument for the use of pressure is shown in figure 8 of this nice study: http://brain.oxfordjournals.org/content/137/3/724
Sorry, I don’t have a spreadsheet for the conversion. For the conversion from filaments to pressure, I measured the bending force of each filament on a balance, divided by filament cross sectional area (from manufacturer), and then fit the plot of pressure vs filament number with a line of best fit. I then used that equation to transform all my data in Graphpad from filament number to pressure. I had to recalibrate periodically, and then just created another conversion equation in Graphpad with the updated values. I mostly used filament number in the SUDO paper to make the comparisons between Chaplan and SUDO easier to follow.
But this might also help - I put a link to a spreadsheet in the comments section of the SUDO article, which I think is lost after Mol Pain was picked up by Sage publications. It was meant to help clarify the adjustment factor and give a means to convert old data collected with Chaplan to SUDO to standardize datasets. here is the sheet: http://bit.ly/1MJhnwd
Thanks @Boninrp. this is very useful. That spreadsheet is what I was looking for in a way. I still will have to determine the final response by eye, but thanks for showing how you converted.
I appreciate your attention to the details of how to analyze and interpret VF data. We had a recent discussion about log representation of sensory testing in light of the Weber-Fechner law:
Do you put your data on a log scale or do a log transformation of the data and then do a linear scale?
Calibration of VF hairs is important too. Thanks for pointing out how you do that. I don’t do it nearly enough. I suppose that even if the actual gram forces are inaccurate, relative differences between the mice at any given time will be valid still. But then if you don’t calibrate, any differences between mice over a period of time (weeks or months) won’t really hold.
I’m going to continue to use SUDO and will encourage others too as well.
Hi @Boninrp
Thanks again for your contributions here and for the SUDO method. VF testing is more complex than we give it credit for.
So, I’m trying to make a pipeline for data analysis. Your spreadsheet was helpful.
I adapted some quick code to find all response combinations:
''' python code to generate all combinations of binary (0,1)'''
#first generate a list
#source: https://stackoverflow.com/questions/14931769/how-to-get-all-combination-of-n-binary-value
import itertools
lst = list(map(list, itertools.product([0, 1], repeat=5)))
# convert to column strings
for i in lst:
print(''.join(map(str, i)))
Output
>>>
00000
00001
00010
00011
00100
00101
00110
00111
01000
01001
01010
01011
01100
01101
01110
01111
10000
10001
10010
10011
10100
10101
10110
10111
11000
11001
11010
11011
11100
11101
11110
11111
>>>
Then I plugged this into your spreadsheet, and got all the final filament values.
For converting to pressure, I was having a hard time finding the cross-sectional area for the VFs I use:
But I did find for this, where I could derive the area, assuming there is correspondence between filament area between manufacturers.
I’m going to follow your lead and use pressures.