I am doing analysis of structural brain networks using BRAPH to compare different groups. Thanks a lot for this useful software!
I would like to perform group comparisons on BUD graphs across different densities. What would be the appropriate method to define the density range?
I read that the density range should include only fully connected and not fragmented graphs (i.e. it should exclude disconnected networks and random topologies). Is there a way to define these range boundaries using BRAPH or matlab?
Thank you for your kind words and all the best when using Braph. Hopefully you will find it useful in your research.
While the density range definition is somewhat arbitrary, you are right, one way of doing it is to include only graphs that are connected. Braph does not give you this information directly, however you can infer this by calculating the path length for different densities. The disconnected nodes will have a path length given as NaN. Therefore, the characteristic path length will not be calculated for these densities. As a result, by checking out the results of the calculation of nodal and global path lengths, you can infer the range of densities that gives you connected graphs.
One alternative way of doing this is to include only networks that have small-world characteristics, since it has been shown that this is the type of network organization present in the healthy brains.
Hope this helps. If there is anything more I can do, please let me know.