help > Design matrix not found or inconsistent error
Showing 1-4 of 4 posts
Display:
Results per page:
Feb 10, 2021  04:02 PM | YaeJi Kim
Design matrix not found or inconsistent error
Dear NBS experts, 

Hello, I am currently woking on identifying subnetworks correlated with behavioral score. 
I have one single group, and just one variable for correlation analysis. 
I set design matrix (n = 240): 
1 0.6
1 -0.4
1 -1.5
1 0.2
1 -0.7 
.  .
.  .
.  . 

so on.. 

And I have connectivity matrices in .txt format. 
I set t-test with t-threshold of 3.0. 

As I clicked 'run NBS' button, the preview panel shows 'Design matrix not found or inconsistent'. 
To resolve the problem I tried few troubleshooting, 
1) Checked the row of the design matrix and the .txt files, and they were all 240. --> checked 
2) I tried schizophreniaexample, and it worked perfectly fine. 
3) remove underscores '_' in file names of connectivity matrices. 

I have no idea what is wrong with this situation, because this is my second time using NBS, and my previous experience with NBS was fantastic. 
Please give me any suggestion to fix this issue. 

Thank you. 

Sincerely, 
Jean
Feb 11, 2021  12:02 AM | Andrew Zalesky
RE: Design matrix not found or inconsistent error
Hi Jean, 

I suggest checking that your connectivity matrices are all valid and can be read. 

The most likely cause for the error is that some of the matrices cannot be read/opened. Make sure that all the matrices contain the same number of columns and rows, and none of the matrices contain empty values or values that are not numbers. 

Another possibility is that the design matrix contains an extra row, which is being interpreted as an additional subject.

If this still does not work, you can send me your design matrix and connectivity matrices and I would be happy to take a look. 

Andrew


Originally posted by YaeJi Kim:
Dear NBS experts, 

Hello, I am currently woking on identifying subnetworks correlated with behavioral score. 
I have one single group, and just one variable for correlation analysis. 
I set design matrix (n = 240): 
1 0.6
1 -0.4
1 -1.5
1 0.2
1 -0.7 
.  .
.  .
.  . 

so on.. 

And I have connectivity matrices in .txt format. 
I set t-test with t-threshold of 3.0. 

As I clicked 'run NBS' button, the preview panel shows 'Design matrix not found or inconsistent'. 
To resolve the problem I tried few troubleshooting, 
1) Checked the row of the design matrix and the .txt files, and they were all 240. --> checked 
2) I tried schizophreniaexample, and it worked perfectly fine. 
3) remove underscores '_' in file names of connectivity matrices. 

I have no idea what is wrong with this situation, because this is my second time using NBS, and my previous experience with NBS was fantastic. 
Please give me any suggestion to fix this issue. 

Thank you. 

Sincerely, 
Jean
Feb 12, 2021  03:02 PM | YaeJi Kim
RE: Design matrix not found or inconsistent error
Dear Andrew, 

Thank you for the advice. I reviewed my connectivity matrices, and I found one matrix was invalid. 

I ran a correlational analysis in a single group to find the subnetwork reflecting correlational relationship associated with the clinical variable. 
However, I couldn't find any significant result, so I am not sure my design matrix and contrast are correct. 


My design matrix is as follows: 
1 0.6
1 -0.4
1 -1.5
1 0.2
1 -0.7
. .
. .
. .

so on..

which first column indicates global mean and second column indicates the clinical variable. 

I set my contrast as [0 1] for positive correlation, and [0 -1] for negative. 
I ran t-test and set t-threshold ranged from 1.0~3.0. 

Is this design and analysis correct for my hypothesis? 



Sicnerely, 
Jean
Feb 12, 2021  11:02 PM | Andrew Zalesky
RE: Design matrix not found or inconsistent error
Hi Jean, 

the design matrix and contrasts appear correct. 

It may simply be that there is no association with the clinical variable, or you need a larger sample size to detect any potential association. 

You could increase alpha to 0.1, to get a sense for this, but I would not put much credibility into findings found at alpha=0.1.  

Andrew
Originally posted by YaeJi Kim:
Dear Andrew, 

Thank you for the advice. I reviewed my connectivity matrices, and I found one matrix was invalid. 

I ran a correlational analysis in a single group to find the subnetwork reflecting correlational relationship associated with the clinical variable. 
However, I couldn't find any significant result, so I am not sure my design matrix and contrast are correct. 


My design matrix is as follows: 
1 0.6
1 -0.4
1 -1.5
1 0.2
1 -0.7
. .
. .
. .

so on..

which first column indicates global mean and second column indicates the clinical variable. 

I set my contrast as [0 1] for positive correlation, and [0 -1] for negative. 
I ran t-test and set t-threshold ranged from 1.0~3.0. 

Is this design and analysis correct for my hypothesis? 



Sicnerely, 
Jean