Dear NBS forum,
Hello everyone, I am a student who is currently trying to learn how to use NBS for my research project. I have two groups (22 in total), and I am attempting to perform a two-sample t-test with two covariates to assess group differences in structural connectivity. However, when I run the analysis, NBS returns an error indicating that the connectivity matrix cannot be found or that the connectivity matrices are inconsistent.
I was wondering whether there might be an issue with my connectivity matrix text file, or whether there is another issue causing the error. I have attached my connectivity matrix list (.mat file), design matrix, and contrast file for reference.
Thank you so much for your time.
A few things to check
- Make sure that a blank line is not present at end of your design file. This can be interpretted as an additional subject.
- Rather than providing a seperate mat file for each subject, you can concatentate the matricies into a 3D array and provide a single file with that 3D array. It may be that some of the mat files are currently not being read.
- Ensure that you can run the NBS on the test dataset. If things work on the test dataset, try to decipher what might differ in your dataset. Your design matrix looks fine. I think the issue is a minor formatting bug or file specficiation.
Best,
Andrew
Originally posted by Jae Young Park:
Dear NBS forum,
Hello everyone, I am a student who is currently trying to learn how to use NBS for my research project. I have two groups (22 in total), and I am attempting to perform a two-sample t-test with two covariates to assess group differences in structural connectivity. However, when I run the analysis, NBS returns an error indicating that the connectivity matrix cannot be found or that the connectivity matrices are inconsistent.
I was wondering whether there might be an issue with my connectivity matrix text file, or whether there is another issue causing the error. I have attached my connectivity matrix list (.mat file), design matrix, and contrast file for reference.
Thank you so much for your time.
Thank you so much for your help. After going through your suggestions, I was able to run the analysis.
Following my previous analysis, I am now running a linear regression analysis on my structural connectivity data and would like to verify my design file. I used the following two contrasts: [0 1 0 0] and [0 -1 0 0] and my design file was structured like the following.
Example:
[Intercepts, predictor variable, covariate 1, covariate 2]
[1, 45, 0.6, -10.18]
[1, 67, 7.8, 6.82]
[1, 32, 8.3, -0.18]
In addition, I was wondering whether there is a way to open the results without providing the Node Coordinates (MNI) and Node Labels files. I was able to obtain a significant result, but at the moment I do not have the corresponding Node Coordinates, so I am unable to see which node was found to be significant.
Thank you again for your help and have a great weekend.
Hi Jae,
the software needs to know the location of all nodes in order to generate a network visualization. Without node locations, it does not know where to place nodes.
A function is avaible in the package to compute node centres from a parcellation.
Alternatively, you can use the matrix of signficiant connections without forming a visualization. Type "global nbs" in the command window after running the software. This will give you access to a structure called nbs, which stores the relevant results.
Andrew
Originally posted by Jae Young Park:
Thank you so much for your help. After going through your suggestions, I was able to run the analysis.
Following my previous analysis, I am now running a linear regression analysis on my structural connectivity data and would like to verify my design file. I used the following two contrasts: [0 1 0 0] and [0 -1 0 0] and my design file was structured like the following.
Example:
[Intercepts, predictor variable, covariate 1, covariate 2]
[1, 45, 0.6, -10.18]
[1, 67, 7.8, 6.82]
[1, 32, 8.3, -0.18]
In addition, I was wondering whether there is a way to open the results without providing the Node Coordinates (MNI) and Node Labels files. I was able to obtain a significant result, but at the moment I do not have the corresponding Node Coordinates, so I am unable to see which node was found to be significant.
Thank you again for your help and have a great weekend.