Dear gPPI experts,

I'm trying to perform PPI analysis with gPPI toolbox. I want to compare two sessions with different conditions in a block-design experiment.
On the basis of 1st level analysis, I used "create_sphere_image" function to extract VOI targeting one region, instead of using the implemented GUI function ("eigenvariate" button). Like this:


create_sphere_image('SPM.mat', [49 49 -9], {'VOI_49_49_-9'},[6])


Then I made P parameters like this:


subject: 'S5'
directory: 'D:\S5 \gPPI'
VOI: 'D:\S5\gPPI\VOI_49_49_-9_mask.nii'
VOImin: 10
ConcatR: 0
Estimate: 1
contrast: 0
extract: 'eig'
Tasks: {'0' 'Task1' 'Task2'}
Weights: []
analysis: 'psy'
method: 'cond'
CompContrasts: 1
Weighted: 0
Contrasts: [1×1 struct]
equalroi: 0
FLmask: 1
Region: 'ROI_49_49_-9'
P.Contrasts is:
Prefix: [1×1 struct] % Left: {'Sn(1)'}, Right: {'Sn(2)'}
left: {'Task1}
right: {'Task2'}
STAT: 'T'
Weighted: 0
MinEvents: 4
name: 'Task1-Task2'


"ppi_checkstruct" seemed went well, but the PPPI(P) failed. Like this:


Log File: S5_PPPI_2_13_2018_8.log
PPPI Version: 13.1.4-17-2014
Parameters used:
Processing subject: S5
VOI file : D:\S5\gPPI\VOI_49_49_-9_mask.nii
Output file will be: S5_ROI_49_49_-9_session#_cond_PPI_regressors.txt
Contrast : No adjustment
Analysis : Psychophysiological Interactions
Extraction : eigenvariate
Tasks : _0_Task1_Task2
Method : Condition Specific
VOI has 58 voxels in 2.200000e+00x2.200000e+00x3.200000e+00 space
VOI has 32 voxels in 2.200000e+00x2.200000e+00x3.200000e+00 space. This is in the the same space as the input data and functional mask.
VOI has 58 voxels in 2.200000e+00x2.200000e+00x3.200000e+00 space
VOI has 32 voxels in 2.200000e+00x2.200000e+00x3.200000e+00 space. This is in the the same space as the input data and functional mask.
Generate Contrast Vectors
Warning: Missing conditions!!! Invalid Contrast
Invalid Contrast
Generate Contrasts
PPI Contrasts were not estimated for some reason.


How should I resolve this problem?
Please let me know if you need more information or any files to help identify the problem.

Best regards,
Takeshi

Hi Takeshi,

I believe that you simple need to change the prefix settings.

Prefix: [1×1 struct] % Left: {'Sn(1) PPI_'}, Right: {'Sn(2) PPI_'}

If this change doesn't work, can you send me your SPM.mat file and the a mat-file version of the parameter structure and I can take a closer look at the issue.

I'm also assuming that your task names are actually Task1 and Task2 or that your actual script replaced these with their actual names. If not, you'll also want to make that change as well.

The key is to get the Prefix setting correct as leaving out PPI_, I think will use the task columns for the contrast.

Best,
Donald

Dear Dr. McLaren,

I'm very glad to hear from you. I applied the revision you suggested, but the same error message came again. I'd like you to confirm the SPM.mat and the mat-file of P parameters.

Best regards,
Takeshi

Here is the P parameters.

Thank you. Takeshi

Hi,

Do you have the SPM.mat file from the PPI analysis? There was no error that the model wasn't estimated, so there should be an associated SPM.mat for the PPI model.

Best,
Donald

Hi Dr. McLaren,

I attached a SPM.mat file of "PPI_ROI_49_49_-9" folder generated by PPPI(P). I'm not sure whether it is the correct one...

Thank you.
Takeshi

Hi Takeshi,

I can't seem to replicate the issue.

If you load your parameter structure and SPM.mat from the PPI analysis, and run the following command:

[vec,left,right] = createVec(P.Contrasts(1).left,P.Contrasts(1).right,SPM,0,P.Contrasts(1).Prefix,[],4)

What happens? It seems to run through to completion.

Best,
Donald

Dear Dr. McLaren,

Thank you for your message. I checked the createVec result by using the P parameters and SPM.mat file obtained through "PPPI(P)" (what I sent to you before, created under "PPI_ROI_49_49_-9" folder). Then the following message appeared:

>> [vec,left,right] = createVec(P.Contrasts(1).left,P.Contrasts(1).right,SPM,0,P.Contrasts(1).Prefix,[],4)
Valid Contrast
vec =
0 1 0 0 -1 0 0 0

left =
cell
'Task1'

right =
cell
'Task2'

When I used the original SPM.mat, "createVec" said Invalid Contrast.
PPPI(P) seemed that it created exact contrast and stored it in SPM.mat.
Please let me suggest what should I go for next. Should I retry by using SPM.mat under "PPI_ROI_49_49_-9" folder?

Best,
Takeshi

I would remove the PPI folder and re-run the analysis.

It's odd that createVec worked on its on, but not from inside the script and that the script generated the contrast even with the errors, but couldn't do the next step.

Best,
Donald