Power264 and other functional atlases

[Philipp Riedel]

Hi Chris,

You mentioned in your user manual that the Power264 and other functional atlases are work in progress.  Do you happen to have a data.table for the Power264 atlas already that you could share? 

Thanks a lot,

Philipp

[Philipp Riedel]

Hi Chris,

Also, I tried to run the script on a subset of destrieux regions that I thought belong to the default mode network.  The resulting custom atlas only contained 14 ROIs.  I was interested in global efficiency within this subnetwork.  This is also one of the reasons I'm interested in the Power264 atlas.

However, when I ran the script on the custom "DMN" atlas I get the following errors:

[1] "Threshold 1/25; group 1; 07:07:41"

Error in group.cols[lobe.nums] : invalid subscript type 'list'

> traceback()

2: make_brainGraph(g, atlas, ...)

1: set_brainGraph_attr(g.tmp, atlas = atlas, modality = modality, 

       weighting = "pearson", threshold = thresholds[j], subject = covars[groups[i], 

           Study.ID[k]], group = groups[i], use.parallel = FALSE, 

       A = A.norm.sub[[j]][, , inds[[i]][k]])

I=j=k=1

Is it because the custom "DMN" atlas does not contain ROIs from all lobes of the brain?  Is there any other way to easily subset the ROIs to get "global" measures for subnetworks?

Thanks,

Philipp

[Chris Watson]

Hi Philipp,

Sorry for not replying to your initial email ; I had meant to over the weekend.

I do have a file for Power264 but it is not yet ready. I have a few
more pressing tasks to do in the short term, so I don't think I could
have it ready until the end of next week maybe.

RE your second email: I am not sure about the error. Does the your
custom atlas's data.table object have a "lobe" column? The "lobe.nums"
variable is created from the following code:

# "DT" is the atlas data.table
vorder <- match(V(g)$name, DT$name)
lobe.nums <- DT[vorder, as.numeric(lobe)]

So, the "vorder" variable is to ensure that we can match the ordering
of vertices in the graph object itself with the order of the vertices
in the data.table. Then "lobe.nums" should simply be a numeric (or
integer) vector in which, e.g.
Frontal --> 1
Parietal --> 2
Temporal --> 3

and so on. For some reason, the "lobe.nums" that was created is
instead a *list*, not a numeric vector. If there is a lobe column in
the data.table, can you make sure that each row contains only a single
*factor* element? You can check with
str(DMN$lobe)

(if "DMN" is the data.table).

If that doesn't reveal anything, then you can send me a "rda" file
with the atlas's data.table, the "covars" table (probably not
necessary), "groups", "A.norm.sub", and "thresholds"? You can send it
to me off-list if you prefer.

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-he...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/brainGraph-help/6f219c15-5d1c-4dc3-8980-f7dde8e497b0%40googlegroups.com.
>

[Philipp Riedel]

Hi Chris,

No worries.  Thanks for your reply.

Concerning the Power atlas... Since I am mostly interested in the global graph measures I quickly created an atlas data.table from the Excel sheet that Power provides on his webpage.  The Excel sheet includes 'index' and 'mni', for 'name' and 'lobe' I just created random vectors.  If you could provide the complete Power Atlas data.table within the next month, that would be great though.

Concerning the error reported in my second post... You were completely right.  Somehow I had converted the atlas data.table to a data.frame while for example adding a column for the subnetworks.  setDT(atlas) solved the issue.

Thanks a lot!

Philipp

> email to brainGr...@googlegroups.com.