vertex_attr_dt Error in get(g$atlas) : invalid first argument

[NedaRR]

Hi. Thank you for this great package. I've successfully run "set_brainGraph_attr" and "dt.G <- rbindlist(lapply(g, graph_attr_dt))". But I have trouble getting the output from "dt.V <- rbindlist(lapply(g, function(x) rbindlist(lapply(x, vertex_attr_dt))))"

Error in get(g$atlas) : invalid first argument g is a large list (7 elements), that is 7 groups I have. I'd appreciate your help in this regard. Thank you.

[cwatson]

Do all of your graphs have a graph-level attribute named atlas? If they do have that attribute, then please make sure the data.table for the atlas is in the search path. That is, you should be able to type in the atlas name and it will print the table in the console.

[NedaRR]

I do have atlas as an attribute. Also when I type dk in the console, prints the table.

[1] "data.table" "data.frame"```

the graph looks like: 
```$`rMDD+naMCI`[[16]]
IGRAPH fc52d26 UN-- 68 1823 -- 
+ attr: Cp (g/n), Lp (g/n), rich (g/x), E.global (g/n), clust.method (g/c), mod (g/n), density (g/n), conn.comp (g/x),
| max.comp (g/n), num.tri (g/n), diameter (g/n), transitivity (g/n), assortativity (g/n), num.hubs (g/n), E.local (g/n),
| vulnerability (g/n), name (v/c), degree (v/n), knn (v/n), Lp (v/n), btwn.cent (v/n), hubs (v/n), ev.cent (v/n), lev.cent
| (v/n), k.core (v/n), transitivity (v/n), E.local (v/n), E.nodal (v/n), vulnerability (v/n), eccentricity (v/n), comm
| (v/n), color.comm (v/c), comp (v/n), color.comp (v/c), circle.layout.comm (v/n), GC (v/n), PC (v/n), z.score (v/n), atlas
| (v/x), betweenness (v/n), eigen_centrality (v/x), btwn (e/n), color.comm (e/c), color.comp (e/c)
+ edges from fc52d26 (vertex names):
 [1] lBSTS--lcMFG  lBSTS--lFUS   lBSTS--lIPL   lBSTS--lITG   lBSTS--liCC   lBSTS--lLOF   lBSTS--lLING  lBSTS--lMOF   lBSTS--lMTG  
[10] lBSTS--lPARH  lBSTS--lpOPER lBSTS--lpORB  lBSTS--lpTRI  lBSTS--lpostC lBSTS--lPCC   lBSTS--lpreC  lBSTS--lPCUN  lBSTS--lrACC 
[19] lBSTS--lrMFG  lBSTS--lSFG   lBSTS--lSPL   lBSTS--lSTG   lBSTS--lSMAR  lBSTS--lFP    lBSTS--lTP    lBSTS--lTT    lBSTS--rBSTS 
+ ... omitted several edges```
and
```> dk
        name  x.mni  y.mni  z.mni      lobe hemi index                              name.full
 1:    lBSTS -56.00 -44.00   5.00  Temporal    L     1 L bank of the superior temporal sulcus
 2:    lcACC  -2.00  21.00  27.00 Cingulate    L     2            L caudal anterior cingulate
 3:    lcMFG -45.00  18.00  46.00   Frontal    L     3          L caudal middle frontal gyrus
 4:     lCUN  -1.00 -82.00  20.00 Occipital    L     4                               L cuneus
 5:     lENT -16.00 -10.00 -29.00  Temporal    L     5                           L entorhinal
 6:     lFUS -24.00 -54.00 -16.00  Temporal    L     6                             L fusiform
 7:     lIPL -47.00 -70.00  31.00  Parietal    L     7             L inferior parietal lobule
 8:     lITG -56.00 -32.00 -24.00  Temporal    L     8              L inferior temporal gyrus
 9:     liCC  -1.00 -48.00  25.00 Cingulate    L     9             L isthmus cingulate cortex
...```.

[cwatson]

If you email me the object g as a .rds file (using the R function saveRDS) I will take a look.

[NedaRR]

I could finally run dt.V, after I specified atlas="dk" in this function: "g <- Map(function(x, y) lapply(x, set_brainGraph_attr, atlas="dk", modality=modality, group=y), g, as.list(groups)) "

However, groups is changed to "Group" only. Instead of specifying my 7 groups. So now the output for dt.G and dt.V is fine, except that is not specifying the groups. looks like this ``` E.nodal vulnerability eccentricity GC PC z.score modality atlas Group 1: 0.7686567 0.0008480104 2 0.6779777 0.6466049 0.33921242 thickness dk Group 2: 0.5298507 -0.0088299828 3 0.2843628 0.2777778 -2.54583111 thickness dk Group 3: 0.7985075 0.0021585720 2 0.6616119 0.6350000 -0.01725592 thickness dk Group 4: 0.7288557 -0.0007649884 3 0.4810010 0.4412071 0.84554004 thickness dk Group 5: 0.6542289 -0.0037893613 3 0.5890061 0.5714286 -0.58591236 thickness dk Group

your help is highly appreciated!

[cwatson]

Hi, I should have inspected the output above more closely. atlas is not listed as a graph-level attribute above, but rather is vertex-level. This is not correct. I am not sure what commands you ran to create the graphs, but you should assign all of your graphs to have the atlas attribute. You will probably also need to assign the following vertex-level attributes, if they are not present: name, lobe, hemi.

[cwatson]

Regarding your latest issue, please provide more information when posting. I do not know the contents of the groups variable so I am not sure why it is "changed".

[NedaRR]

Atlas is now correctly attributed to group

| (g/n), diameter (g/n), transitivity (g/n), assortativity (g/n), _atlas (g/c_), version (g/x), modality (g/c), Group (g/c), assortativity.lobe
| (g/n), assortativity.lobe.hemi (g/n), asymm (g/n), spatial.dist (g/n), num.hubs (g/n), E.local (g/n), vulnerability (g/n), name (v/c), degree
| (v/n), lobe (v/c), lobe.hemi (v/n), hemi (v/c), x.mni (v/n), x (v/n), y.mni (v/n), y (v/n), z.mni (v/n), z (v/n), color.lobe (v/c),
| circle.layout (v/n), asymm (v/n), dist (v/n), dist.strength (v/n), knn (v/n), Lp (v/n), btwn.cent (v/n), hubs (v/n), ev.cent (v/n), lev.cent
| (v/n), k.core (v/n), transitivity (v/n), E.local (v/n), E.nodal (v/n), vulnerability (v/n), eccentricity (v/n), comm (v/n), color.comm (v/c),
| comp (v/n), color.comp (v/c), circle.layout.comm (v/n), GC (v/n), PC (v/n), z.score (v/n), color.lobe (e/c), dist (e/n), btwn (e/n),
| color.comm (e/c), color.comp (e/c)

I created g like this :

myResids<- get.resid(lhrh, covars=covars, exclude=c("Study.ID", "Group"), method = "sep.groups")
corrs <- corr.matrix(myResids, densities=densities)
g<-lapply(corrs, function(x)
  apply(x$r.thresh, 3,
        graph_from_adjacency_matrix, mode='undirected', diag=F))

covars is

                   Study.ID  Group Age Sex EduCateg
  1:  PACTMD_CMH_0100002_01  aMCI  68   0        6
  2:  PACTMD_CMH_0100010_01  aMCI  71   0        6
  3:  PACTMD_CMH_0100012_01  aMCI  60   1        5
  4:  PACTMD_CMH_0100027_01  aMCI  70   1        4
  5:  PACTMD_CMH_0100031_01  aMCI  85   0        6
 ---                                              
329: PACTMD_CMH_H3060188_01    HC  60   1        6
330: PACTMD_CMH_H3070065_01    HC  67   0        7
331: PACTMD_CMH_H3070076_01    HC  72   1        7
332: PACTMD_CMH_H3070083_01    HC  75   0        5
333: PACTMD_CMH_H3070115_01    HC  69   1        5

and Group is "aMCI, HC, AD, and 4 other groups".

g output is a large list of 7 elements, and looks like this for one group:

$AD[[1]]
IGRAPH 2edeab2 UN-- 68 1139 -- 
+ attr: Cp (g/n), Lp (g/n), rich (g/x), E.global (g/n), clust.method (g/c), mod (g/n), density (g/n), conn.comp (g/x), max.comp
| (g/n), num.tri (g/n), diameter (g/n), transitivity (g/n), assortativity (g/n), atlas (g/c), version (g/x), modality (g/c),
| Group (g/c), assortativity.lobe (g/n), assortativity.lobe.hemi (g/n), asymm (g/n), spatial.dist (g/n), num.hubs (g/n),
| E.local (g/n), vulnerability (g/n), name (v/c), degree (v/n), lobe (v/c), lobe.hemi (v/n), hemi (v/c), x.mni (v/n), x (v/n),
| y.mni (v/n), y (v/n), z.mni (v/n), z (v/n), color.lobe (v/c), circle.layout (v/n), asymm (v/n), dist (v/n), dist.strength
| (v/n), knn (v/n), Lp (v/n), btwn.cent (v/n), hubs (v/n), ev.cent (v/n), lev.cent (v/n), k.core (v/n), transitivity (v/n),
| E.local (v/n), E.nodal (v/n), vulnerability (v/n), eccentricity (v/n), comm (v/n), color.comm (v/c), comp (v/n), color.comp
| (v/c), circle.layout.comm (v/n), GC (v/n), PC (v/n), z.score (v/n), color.lobe (e/c), dist (e/n), btwn (e/n), color.comm
| (e/c), color.comp (e/c)
+ edges from 2edeab2 (vertex names):

$AD[[2]]
IGRAPH 425f85a UN-- 68 1185 -- 
+ attr: Cp (g/n), Lp (g/n), rich (g/x), E.global (g/n), clust.method (g/c), mod (g/n), density (g/n), conn.comp (g/x), max.comp
| (g/n), num.tri (g/n), diameter (g/n), transitivity (g/n), assortativity (g/n), atlas (g/c), version (g/x), modality (g/c),
| Group (g/c), assortativity.lobe (g/n), assortativity.lobe.hemi (g/n), asymm (g/n), spatial.dist (g/n), num.hubs (g/n),
| E.local (g/n), vulnerability (g/n), name (v/c), degree (v/n), lobe (v/c), lobe.hemi (v/n), hemi (v/c), x.mni (v/n), x (v/n),
| y.mni (v/n), y (v/n), z.mni (v/n), z (v/n), color.lobe (v/c), circle.layout (v/n), asymm (v/n), dist (v/n), dist.strength
| (v/n), knn (v/n), Lp (v/n), btwn.cent (v/n), hubs (v/n), ev.cent (v/n), lev.cent (v/n), k.core (v/n), transitivity (v/n),
| E.local (v/n), E.nodal (v/n), vulnerability (v/n), eccentricity (v/n), comm (v/n), color.comm (v/c), comp (v/n), color.comp
| (v/c), circle.layout.comm (v/n), GC (v/n), PC (v/n), z.score (v/n), color.lobe (e/c), dist (e/n), btwn (e/n), color.comm
| (e/c), color.comp (e/c)
+ edges from 425f85a (vertex names):

[cwatson]

I meant the value of the vector groups. Please paste the actual value of that variable.

[NedaRR]

groups [1] "Group"

[NedaRR]

One other question I got, how is "hub" been defined/calculated in the Braingraph? what does hub=0, hub=1, and hub=2 and more represent? Is hub referring to hub betweenness? I am following this:

# Count vertices with betweenness > mean + sd (i.e. 'hub' regions)
dt.Vhc[density == round(densities[1], 2) & hubs > 0, .N, by=Group] 

I am interested in the hub betweenness > 2 sd

[cwatson]

Regarding the groups variable, this must be a vector, with length equal to the number of your groups (i.e., 7), in which each element is the group name/label.

Regarding hubs, please see the reference in the help for the hubness function. As it states, the value for hubs is from 0 to 4, depending on how many times the vertex was in the top 20% for one of the 4 metrics. The authors of the referenced paper choose a cutoff value of 2 or higher to be considered a hub. So, it is not hub betweenness and thus does not have a meaningful standard deviation.

[NedaRR]

Thank you so much! Its working perfectly now!

[cwatson]

I'm glad to hear it. I will close the issue.