Re: [Icatb-discuss] FIT: How does the tool accept correlation matrix?

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

Hi Srinivas,

That change solved the error. Thank you very much.
Also, it seems that the current batch files now can accept ‘eeg’, ‘fmri and ‘smri’. If you include more modalities there, it could be more convenient.

Thanks,
Kai

From: Srinivas Rachakonda <sra...@mr...<mailto:sra...@mr...>>
Sent: 2018年11月6日 23:03
To: Kai Wang <Kai...@Co...<mailto:Kai...@Co...>>
Cc: Vince Calhoun <vca...@mr...<mailto:vca...@mr...>>
Subject: Re: [Icatb-discuss] FIT: How does the tool accept correlation matrix?

Thanks Kai. You can change the number of principal components from 196 to a number less than 136 which is the least dimension of the fnc data-set. Let me know if you have questions.

On Tue, Nov 6, 2018 at 4:02 PM Kai Wang <Kai...@co...<mailto:Kai...@co...>> wrote:
Hi Srinivas,

Here they are,
https://drive.google.com/file/d/1thjJD4L2Zjn6EefYmcbauZTmybGmSiMy/view?usp=sharing

Thanks,
Kai

From: Srinivas Rachakonda <sra...@mr...<mailto:sra...@mr...>>
Sent: 2018年11月6日 15:21
To: Kai Wang <Kai...@Co...<mailto:Kai...@Co...>>
Cc: Vince Calhoun <vca...@mr...<mailto:vca...@mr...>>
Subject: Re: [Icatb-discuss] FIT: How does the tool accept correlation matrix?

Thanks Kai. Could you upload all the ascii files of fnc correlations?

On Tue, Nov 6, 2018 at 10:40 AM Kai Wang <Kai...@co...<mailto:Kai...@co...>> wrote:
Hi Srinivas,

It’s here. Thank you for your effect.
https://drive.google.com/file/d/1psqFk2CTzK5i7MF36Po-7NzbeRixiD2u/view?usp=sharing

Kai

From: Srinivas Rachakonda <sra...@mr...<mailto:sra...@mr...>>
Sent: 2018年11月6日 9:08
To: Kai Wang <Kai...@Co...<mailto:Kai...@Co...>>
Cc: Vince Calhoun <vca...@mr...<mailto:vca...@mr...>>
Subject: Re: [Icatb-discuss] FIT: How does the tool accept correlation matrix?

Could you share subject fnc correlations in dropbox or google drive? Also please send the fusion info file (*fusion*mat).

On Mon, Nov 5, 2018 at 2:12 PM Srinivas Rachakonda <sra...@mr...<mailto:sra...@mr...>> wrote:
Hi Kai,

You could save the fnc matrix as behavioral. Do you have fmri as one of the features or you were trying to save the fnc matrix in nifti format?

Thanks,
Sri

On Mon, Nov 5, 2018 at 1:22 PM Kai Wang <Kai...@co...<mailto:Kai...@co...>> wrote:
Thank you, Srinivas, for the quick reply.

Yes, I save one file for each subject (ends in “.asc”) as in the examples, with matlab’s “save” function (with “-ascii”). The vector contains with only the lower triangle portion excluding cells on the diagonal. Yet I still get those errors…
         Thank you for adding a ‘FNC’ modality. We can wait a few days. We have about 200 subjects.

Thanks,
Kai

From: Srinivas Rachakonda <sra...@mr...<mailto:sra...@mr...>>
Sent: 2018年11月5日 12:03
To: Kai Wang <Kai...@Co...<mailto:Kai...@Co...>>
Cc: Vince Calhoun <vca...@mr...<mailto:vca...@mr...>>
Subject: Re: [Icatb-discuss] FIT: How does the tool accept correlation matrix?

Hi Kai Wang,

It needs to be saved as a vector in a ascii file one per subject. Usually lower triangular portion needs to be selected as the rest of the elements are not of use. You can select behavioral as the modality type. If you can wait until Friday this week I can add  FNC as a feature. I am assuming you have multiple subjects in your analysis.

Thanks,
Sri

---------- Forwarded message ---------
From: Kai Wang <Kai...@co...<mailto:Kai...@co...>>
Date: Sun, Nov 4, 2018, 11:20 PM
Subject: Re: [Icatb-discuss] FIT: How does the tool accept correlation matrix?
To: ica...@li...<mailto:ica...@li...> <ica...@li...<mailto:ica...@li...>>

Thank you, Vince, for the advice.

However, I tried both ways and still didn't get the job through. Please see the error messages below for each type of trial:
a. resting-state correlation matrix saved as a vector in asc2 format and a modality of "eeg" was selected.
b. the matrix saved as a vector in asc2 format and a modality of "gene" was selected.
c. the matrix saved as a vector in asc2 format and a modality of "behavioral" was selected.
d. the matrix was transformed to nifty format (dimensions: 136 by 1 by 1) with fsl command "save_avw". A modality of "fmri" was selected.
e. the matrix was transformed to nifty format (dimensions: 136 by 136 by 136) where the vector was save in the first column and all other cells were set to zero. A modality of "fmri" was selected.

We wanted to deal with the resting state modality pretty much like the way you did it in the following paper.
Lerman-Sinkoff, D. B., Sui, J., Rachakonda, S., Kandala, S., Calhoun, V. D., & Barch, D. M. (2017). Multimodal neural correlates of cognitive control in the Human Connectome Project. NeuroImage, 163, 41-54.

Thank you very much,
Kai Wang
Postdoc at University of Colorado Boulder

## error log
a. eeg
...
Loading data for read & nback & area & thickness & lgi & rest ...