help > NBSview very slow/failing
Showing 1-6 of 6 posts
Display:
Results per page:
Nov 6, 2018  04:11 AM | Daniel Brennan - Icahn School of Medicine at Mount Sinai; New York University
NBSview very slow/failing
Hi everyone, 
I'm having trouble running NBSview; its taking an extremely long time to load, if at all, even if simply feeding NBSview a previously-computed network. Anyone ever run into this? Is it an issue with MATLAB? Since running into this problem I've re-installed everything I can imagine, to no avail. Any help is appreciated!
Nov 6, 2018  10:11 AM | Andrew Zalesky
RE: NBSview very slow/failing
Hi Daniel,

NBSview should be quick, unless the network that you are seeking to view is particularly large.

How many nodes and edges does the network comprise?

Anything up to 500 or so nodes should be feasible.

Andrew



Originally posted by Daniel Brennan:
Hi everyone, 
I'm having trouble running NBSview; its taking an extremely long time to load, if at all, even if simply feeding NBSview a previously-computed network. Anyone ever run into this? Is it an issue with MATLAB? Since running into this problem I've re-installed everything I can imagine, to no avail. Any help is appreciated!
Nov 6, 2018  03:11 PM | Daniel Brennan - Icahn School of Medicine at Mount Sinai; New York University
RE: NBSview very slow/failing
Hi Andrew,
Thanks for your reply. My network is 470 nodes, so well within range, and was working sufficiently until a day ago. Is there some memory that's stored every time NBS is run that can be cleared?
Nov 6, 2018  10:11 PM | Andrew Zalesky
RE: NBSview very slow/failing
470 nodes is relatively large, particularly if the edge density is high. Network with this number of nodes can comprise >100,000 edges.

A history file (.mat) is stored each time the software is run, however, this should not affect NBSView.

Make sure that the network you are trying to view has a reasonable edge density.

Originally posted by Daniel Brennan:
Hi Andrew,
Thanks for your reply. My network is 470 nodes, so well within range, and was working sufficiently until a day ago. Is there some memory that's stored every time NBS is run that can be cleared?
Nov 6, 2018  11:11 PM | Daniel Brennan - Icahn School of Medicine at Mount Sinai; New York University
RE: NBSview very slow/failing
Thanks Andrew, I think it might be that the graph density is too high. Regardless, how can one find/clear the history.mat files? Also, is there a way to salvage the results of an NBSrun if the results do not meet alpha threshold? Lets say I run 5000 perms and get a result of 0.011, and my alpha was 0.01; is that network salvageable? 

Thanks again
Nov 6, 2018  11:11 PM | Andrew Zalesky
RE: NBSview very slow/failing
history.mat is stored in the NBS1.2 directory. This file can be simply deleted.

Simply change the alpha threshold and rerun (i.e. hit the nbs_connectome button). It will rerun quicker because the permutations will have been stored from the first run.






Originally posted by Daniel Brennan:
Thanks Andrew, I think it might be that the graph density is too high. Regardless, how can one find/clear the history.mat files? Also, is there a way to salvage the results of an NBSrun if the results do not meet alpha threshold? Lets say I run 5000 perms and get a result of 0.011, and my alpha was 0.01; is that network salvageable? 

Thanks again