help
help > RE: directed connectivity
Dec 3, 2021 11:12 PM | Andrew Zalesky
RE: directed connectivity
Hi Yolanda,
Both dNBS and NBS will take into account the sign of edges in the same way. The key difference is that dNBS allows for directed edges, whereas NBS does not.
Using a t-test on the difference between T2 and T2 sounds reasonable. The statistical framework is the same for NBS and dNBS. You may want to reach out to the developer of dNBS, Max von Gellhorn (vongellhorn@gmx.net)
best,
Andrew
Originally posted by Yolanda Schlumpf:
Both dNBS and NBS will take into account the sign of edges in the same way. The key difference is that dNBS allows for directed edges, whereas NBS does not.
Using a t-test on the difference between T2 and T2 sounds reasonable. The statistical framework is the same for NBS and dNBS. You may want to reach out to the developer of dNBS, Max von Gellhorn (vongellhorn@gmx.net)
best,
Andrew
Originally posted by Yolanda Schlumpf:
Dear Andrew
Thank you for your respond! O.k. I see there was a misunderstanding. I thought dNBS calculates directionality. Meanwhile I managed to calculate directed/asymmetric connectivity matrices in another tool.
I have a 2 (Groups; patients and controls) x 2 (Time points, T1 and T2) factorial design. I calculated connectivity matrices for both groups and both measurement points. I'm interested in the interaction effect group x time point. Can I run a t-test on the difference connectivity matrices (i.e., [Patient T2 - Patient T1] vs. [Controls T2 - Controls T1])? I know that this is possible for an undirected connectivity analysis in NBS. But I'm wondering if it produces correct results in the directed connectivity analysis as well?
Do I understand it correctly that the mathematical procedure in dNBS is the same as in NBS unless that dNBS takes into account the sign. of the edges (+/- t-values)?
Kind regards,
Yolanda
Thank you for your respond! O.k. I see there was a misunderstanding. I thought dNBS calculates directionality. Meanwhile I managed to calculate directed/asymmetric connectivity matrices in another tool.
I have a 2 (Groups; patients and controls) x 2 (Time points, T1 and T2) factorial design. I calculated connectivity matrices for both groups and both measurement points. I'm interested in the interaction effect group x time point. Can I run a t-test on the difference connectivity matrices (i.e., [Patient T2 - Patient T1] vs. [Controls T2 - Controls T1])? I know that this is possible for an undirected connectivity analysis in NBS. But I'm wondering if it produces correct results in the directed connectivity analysis as well?
Do I understand it correctly that the mathematical procedure in dNBS is the same as in NBS unless that dNBS takes into account the sign. of the edges (+/- t-values)?
Kind regards,
Yolanda
Threaded View
| Title | Author | Date |
|---|---|---|
| Yolanda Schlumpf | Nov 29, 2021 | |
| Andrew Zalesky | Nov 29, 2021 | |
| Yolanda Schlumpf | Dec 2, 2021 | |
| Andrew Zalesky | Dec 3, 2021 | |
| Yolanda Schlumpf | Dec 4, 2021 | |