I was trying to use alphaSim to do a one-dimension Monte Carlo simulation to estimate the significance of my one-dimension ERP data. However, the AlphaSim is no longer available in the lastest version of AFNI, so I used 3dClustSim instead. But I found that 3dClustSim required at least 256 data point to run, but my timeseries ERP data only have 76 data points, so it didn’t work and gave me an error message. My advisor told me that she used to do the same thing with the function “AlphaSim”, and it didn’t have the error message, which means that “AlphaSim” doesn’t have the 256 data points limitation. So I want to install AlphaSim in my AFNI(which is the latest version), but I don’t know how to do that. Can I install the function directly, or I must download an older version of AFNI?
I was just thinking about this the other day!
You can get AlphaSim from an older binary or compile AlphaSim from source. What operating system are you running AFNI on?
I am using Windows, and I installed a Windows Subsystem for Linux to run AFNI. In my situation, how can I download the older binary including AlphaSim? Is there a command to do so?
Are those 76 data points in the time domain? If so,
AlphaSim is meant to handle 3-dimensional datasets, and is probably not an appropriate tool for the case. One option is use to
3dFDR with its option
-input1D. Or, simply present the result as is.
I also wonder if the ERP data has the same 3D structure that is appropriate for clusterizing? I think the assumptions of most clustering are more consistent with more ball-like, 3D shapes, as opposed to surface ones.
SurfClust would be more appropriate for 2D surfaces.
It is more like the guassian filter they used in the simulation is a good fit for our 1d question, and we treat it as an 1d simulation, and our lab used it in the past. I am not sure if the 2d function is a better choice, but I will look into the manual. Thanks for the advice!