Transform HCP structural connectivity to TVB

[Silvana Huertas]

Dear, 

I'm writing because I'm creating some models using TVB (scripting). What I want to do, is to use the structural and functional connectivity obtained from the HCP (https://search.kg.ebrains.eu/instances/88507924-8509-419f-8900-109accf1414b) and input them into TVB. However, I think I need to do some transformations to these data to be able to use it in TVB. Do you know which transformations I need to perform? 

I'm doing region base modeling. I already have the code I created for the model but for now I'm using the predefine data of TVB, but I want to use the one from HCP. 

If necessary, I can send the code I created for my model to you. 

Thank you in advance for your help,

Silvana

[Michael Schirner]

Hi Silvana,

there are several ways to import data to TVB. I would think the easiest at the moment is probably to Import Connectivity from Zip, respectively the Connectivity.from_file() function. Since the SC weights files already contain the keyword `weight` I expect no change will be needed for them before zipping. For the `_distances.tsv` files I think it will be necessary to rename them to contain the keyword `tract`. Note that no "centres", "areas", "orientations", etc. are part of the HCP data set, but also not strictly needed for pure region-level simulation without e.g. EEG forward modelling etc. In case the importer complains about such files missing a workaround can be to impute random data with the format expected by importer.

Best,

Micha

[Silvana Huertas]

Hi Michael, 

Thank you for your quick answer. 

I need to perform EEG forward modelling. Do you know if there is a way to obtain the centres, areas, orientations etc from the HCP dataset to do the forward modelling?

Thank you for your help, 

Silvana Huertas

[Michael Schirner (michael.schirner@bih-charite.de)]

Hi Silvana,

the shared dataset on EBRAINS contains no information about centres, etc. it only contains region-by-region (aggregated) SC, FC and region-average fMRI time series. To create forward models for single subjects from their cortical surfaces a complex process is necessary (we describe it briefly here and here; the tutorials of the Brainstorm toolbox here are valuable). 

Since the used parcellation (Glasser / HCP) is becoming increasingly common for brain network models it could be that other researchers already created an averaged forward model for these HCP subjects that can be shared.

Best,

Micha

[Silvana Huertas]

Hi Michael, 

Thank you for your help. I will take a look at the links you send me and check if maybe someone else has already created an average forward model for these subjects. 

I also had another question. For my model I use the rm = region_mapping.RegionMapping.from_file. Is the region mapping file available in the data set of HCP that we are talking about? 

Thanks, 

Silvana

[Michael Schirner (michael.schirner@bih-charite.de)]

Yes.

[Silvana Huertas]

Thanks :) 

Have a great week