Group analysis - Time generalization analysis

[Alireza Karami]

Hello

I want to do a group analysis on the results from time generalization analysis using TFCE and find the significance cluster and reflects the clusters using outline, similar to the figure below:

I did multiple-comparison correction for a time course before but I do not know how to do it on matrix. Would you please help me to do the group analysis in CoSMoMVPA?

[Nick Oosterhof]

Greetings,

> On 9 May 2023, at 14:48, Alireza Karami <alire...@gmail.com> wrote:
>
> I want to do a group analysis on the results from time generalization analysis using TFCE and find the significance cluster and reflects the clusters using outline, similar to the figure below:

> <Screenshot 2023-05-09 144041.jpg>
>
> I did multiple-comparison correction for a time course before but I do not know how to do it on matrix. Would you please help me to do the group analysis in CoSMoMVPA?

I have added an entry to the FAQ:

https://www.cosmomvpa.org/faq.html#run-group-analysis-on-time-generalization-results

[Alireza Karami]

Thank you, Nik! But I have another problem. I do not have the result from the "cosmo_dim_generalization_measure". What I currently have is 2D matrices (time generalization matrices) from "cosmo_unflatten". Is there any way to implement the group analysis on the matrix? Or is it possible to convert the matrix to the format of "cosmo_dim_generalization_measure" function's result?

[Nick Oosterhof]

> On 28 May 2023, at 23:55, Alireza Karami <alire...@gmail.com> wrote:
>
> Thank you, Nik! But I have another problem. I do not have the result from the "cosmo_dim_generalization_measure". What I currently have is 2D matrices (time generalization matrices) from "cosmo_unflatten". Is there any way to implement the group analysis on the matrix? Or is it possible to convert the matrix to the format of "cosmo_dim_generalization_measure" function's result?

cosmo_flatten and cosmo_unflatten are more or less inverse functions, so you could use cosmo_flatten I guess. Out of curiosity, if you have 2D matrices from cosmo_unflatten, how did you create the input used in cosmo_unflatten if not from cosmo_dim_generalization_measure?

[Alireza Karami]

Thank you, Nik!

Yes, I used cosmo_dim_generalization_measure but then I used cosmo_unflatten to get 2D matrices and I saved them. Now, I only have 2D matrices and I do not want to redo the analysis again so I was seeking a way to convert the 2D matrices to the flat format.

[Nick Oosterhof]

> On 4 Jun 2023, at 16:17, Alireza Karami <alire...@gmail.com> wrote:
>
> Yes, I used cosmo_dim_generalization_measure but then I used cosmo_unflatten to get 2D matrices and I saved them. Now, I only have 2D matrices and I do not want to redo the analysis again so I was seeking a way to convert the 2D matrices to the flat format.

Well, if you are able to recreate dim_labels (which should be easy) and dim_values (the time points) then you should be able to use cosmo_flatten.

As general advice: be aware that you are losing a bit of information when only storing the 2D matrices instead of the dataset structures.