MST and future updates

[neurog...@gmail.com]

Hi Chris,

 

I have recently found your software tool from your post to the freesurfer mailing list and I have found it very interesting!

 

One thing holding me back from utilising BraiinGraph in my own work is the lack (from what I can see) of functionality regarding the minimum spanning tree. The MST approach allows you to balance the need for a fully connected network, as well as maintaining relatively low thresholds to keep the networks biologically meaningful (and non-random). This is usually achieved by allowing the generation of the MST then allowing the addition of the top percentage of connections ON TOP of this MST network.

 

This would then allow you to implement metrics such as diameter, leaf number, eccentricity and hierarchy in relation to the MST (Stam et al., 2014).

 

Is there any plans to implement these approaches for the Volumetric and DTI networks outlined in your software?

 

Cheers,

B.

[Chris Watson]

Hi, although I do not have anything specific for dealing with MST's, there is already an igraph function, aptly named "mst". This function returns the MST in the form of an igraph graph object. If you input a graph with certain attributes (after running "set_brainGraph_attr", for example) then it will retain those attributes. So the MST would retain (x,y,z) coordinates, vertex name, lobe membership, etc. Then you can run, e.g. "diameter" or "eccentricity" on that subgraph and any of a number of other functions.

I would like to implement something specific but I don't know when I will get around to it. If you have a specific use case in mind, please let me know and I can try to implement it.

Chris

--
You received this message because you are subscribed to the Google Groups "brainGraph-help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to brainGraph-help+unsubscribe@googlegroups.com.
To post to this group, send email to brainGraph-help@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/brainGraph-help/85b6b4c5-fc31-46bb-8e73-452cd398621b%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Daniel J. King]

Hi Chris,

This is definitely something that I would love to see too as part of the braingraph capabilities. Whilst iGrah can generate the MST I would want a similar function to what the poster above was suggesting. I would want to be able to, instead of calculating grah attributes over densities, I would want to be able to calculate them for the graphs 1..i where each graph represents the MST PLUS the highest density connections. for example MST+10% strongest connections... MST+30% greatest connections. This would ensure that all attributes were calculated on fully connected graphs and thus are more ecologically valid.

Whilst I could generate the MST and probably pass the greatest strength connections my worry would be trying to pass those graph objects down to other functions, such as the permutation testing etc.

WOuld this be something you could implement?

BW

Dan

On Wednesday, May 10, 2017 at 12:08:39 AM UTC+1, Chris Watson wrote:

Hi, although I do not have anything specific for dealing with MST's, there is already an igraph function, aptly named "mst". This function returns the MST in the form of an igraph graph object. If you input a graph with certain attributes (after running "set_brainGraph_attr", for example) then it will retain those attributes. So the MST would retain (x,y,z) coordinates, vertex name, lobe membership, etc. Then you can run, e.g. "diameter" or "eccentricity" on that subgraph and any of a number of other functions.

I would like to implement something specific but I don't know when I will get around to it. If you have a specific use case in mind, please let me know and I can try to implement it.

Chris

On Tue, May 9, 2017 at 3:28 PM, <neurog...@gmail.com> wrote:

Hi Chris,

 

I have recently found your software tool from your post to the freesurfer mailing list and I have found it very interesting!

 

One thing holding me back from utilising BraiinGraph in my own work is the lack (from what I can see) of functionality regarding the minimum spanning tree. The MST approach allows you to balance the need for a fully connected network, as well as maintaining relatively low thresholds to keep the networks biologically meaningful (and non-random). This is usually achieved by allowing the generation of the MST then allowing the addition of the top percentage of connections ON TOP of this MST network.

 

This would then allow you to implement metrics such as diameter, leaf number, eccentricity and hierarchy in relation to the MST (Stam et al., 2014).

 

Is there any plans to implement these approaches for the Volumetric and DTI networks outlined in your software?

 

Cheers,

B.

--
You received this message because you are subscribed to the Google Groups "brainGraph-help" group.

To unsubscribe from this group and stop receiving emails from it, send an email to brainGraph-he...@googlegroups.com.
To post to this group, send email to brainGr...@googlegroups.com.

[Chris Watson]

Hi Dan,

Yes I am interested in implementing this as well. However, it is not currently at the top of my to-do list, so I can't give an accurate timeline for how long that would take. I will certainly start thinking about it and writing/testing some initial code.

Chris

To unsubscribe from this group and stop receiving emails from it, send an email to brainGraph-help+unsubscribe@googlegroups.com.
To post to this group, send email to brainGraph-help@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/brainGraph-help/6561da0b-3625-4b21-befb-c0e3f1e85265%40googlegroups.com.

[Daniel J. King]

Hi Chris, 

I completely understand. I look forward to seeing how it goes in the future!

BW

Dan

[Daniel J. King]

Hi Chris,

I was just interested in wether or not you have had a moment to look at this at all since we last spoke. I recently attempted this myself with the Igraph package but reached a deadend on trying to implement such capabilities. I only ask as I have some interesting findings which I think may be explainable by disconnection of graphs but cannot test this without the MST.

Many thanks,

Dan

[Chris Watson]

Hi Dan, I have not wholly forgotten this request but admittedly haven't made any progress on it. I have been working extensively on bringing functionality from the "mediation" package to work with my package, and writing a chapter in the User Guide to go along with it. I am essentially done with this (I have a working function and have run several of my own analyses) and plan on including it in an upcoming version very soon.

But I was actually thinking again about the MST this past week, and it is either #1 or #2 on what I would like to tackle next. I will look at it more in-depth this weekend. But thanks for bumping up this issue, I am eager to work on it.

Chris

To unsubscribe from this group and stop receiving emails from it, send an email to brainGraph-help+unsubscribe@googlegroups.com.
To post to this group, send email to brainGraph-help@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/brainGraph-help/1f6f800a-ba7d-4684-92c4-a001cff7c204%40googlegroups.com.

[Daniel J. King]

Hi Chris,

Thanks for keeping me updated, very much appreciate it :) Looking forward to seeing the functionality of the mediation analysis too!