problems with g, GC.wt and PC.wt.

[Virva Saunavaara]

Hi Chris,

Thank you for the wonderful analysis package! 

System info:

platform       x86_64-apple-darwin17.0  

version.string R version 4.0.3 (2020-10-10)

RStudio Version 1.2.5033

I have tried this with brainGraph version 3.0.0 and 3.0.1. 

I’m working with vertex measures GC.wt and PC.wt. My problem is that the values of the parameters are the same. Values GC and PC are not the same. They (GC.wt and PC.wt) were not the same in version 2.7.3. 

I also had problems when I created the g and the g.group. Maybe this is the cause of my additional problems?

As a start I used the example code on manual page 46: 

g[[j]] <- make_brainGraphList(A.norm.sub[[j]], atlas, modality=modality, weighting='sld', threshold=thresholds[j], 

weighted=TRUE, gnames=covars.dti$Study.ID, groups=covars.dti$Group) 

I get an error: Error in { : task 1 failed - "unused argument (groups = c(2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1))"

and the rds-files are not saved.

I made some changes to the code. I attached the code to this mail. Shortly:

For g[[j]]: “groups=covars.dti$Group” was changed to “grpNames=covars.dti$Group”

and 

For g.group[[j]]: I added “grpNames=grps” to the end. 

After this I write the values to data tables. I used the following code (manual page 47):

dt.V <- rbindlist(lapply(g, vertex_attr_dt))

dt.G <- rbindlist(lapply(g, graph_attr_dt))

dt.V.group <- rbindlist(lapply(g.group, vertex_attr_dt)) 

dt.G.group <- rbindlist(lapply(g.group, graph_attr_dt))

If I open the dt.V data table, the GC.wt and PC.wt have the same values. 

At the later stage of the analysis, I try to do the vertex-wise group analysis using g. I get the same result nodes and values for GC.wt and PC.wt. 

I can’t figure out what I did wrong. I hope you can help me. 

With Best Regards,

Virva

[Chris Watson]

I don't know without seeing the data, but neither of those functions have changed in a long time. You can try calculating each for a single graph and seeing if the numbers differ. E.g.

GC1 <- gateway_coeff(g[[1]][1], V(g[[1]][1])$comm.wt)

PC1 <- part_coeff(g[[1]][1], V(g[[1]][1])$comm.wt)

If they are different from what you see in your data tables, then there could be a bug when calculating from within another function.

--
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/CAGs%2BzKt1xB5qpD9yJqaiA2yPC5fVHxgLnoemyV9iKfi%3D2ExNgQ%40mail.gmail.com.

[virva.sa...@gmail.com]

Hi,

Thank you for your kind reply! 

The separate calculations give different values for GC.wt and PC.wt. Alto the values are quite different what I get from dt.V. I attached the notebook html to this message. 

Best Regards,

Virva

[Chris Watson]

How about instead, you print the vertex attributes directly; i.e.,

V(g[[1]][1])$GC.wt

V(g[[1]][1])$PC.wt

I suspect these will be the same but would like to confirm.

To view this discussion on the web visit https://groups.google.com/d/msgid/brainGraph-help/8e362ad1-e6b7-44d3-b05f-e69009ab3f1dn%40googlegroups.com.

[Virva Saunavaara]

Hi,

They look the same. See the attached note book. New lines at the end.

Thanks again!

With Best Regards,

Virva

To view this discussion on the web visit https://groups.google.com/d/msgid/brainGraph-help/CAHVz74iPo23_qsNQ0r1haY%3Dx6vAHngDaeU%3DAe2rGJ0gDoqKtfg%40mail.gmail.com.

[Chris Watson]

Then if you'd like me to find out where the bug is occurring, I will need to see your data. Please send all data I would need to reproduce the issue; this will likely be the 3D array of connectivity matrices (if these are single-subject graphs) in addition to the covariates table. Please also send me the code you used to generate your data and results; I would prefer just a simple R script, but if you only have Rmd that is ok.

Chris

P.S. It would be much simpler to use the "data.table" syntax without creating a function like "foo" which just complicates things. So you would write:

DT[Study.ID == '153' & threshold == 0.01, .(Study.ID, threshold, GC.wt, PC.wt)]

To view this discussion on the web visit https://groups.google.com/d/msgid/brainGraph-help/CAGs%2BzKvMAMmtYzBOJwf9XNU5cEpB20ch_XE%3DWM4Ymacr_LHf8g%40mail.gmail.com.

[virva.sa...@gmail.com]

Hi,

You can find the data attached to this mail. It is only small part of the full data set. These are the original matrix and additional data. I hope this format is ok, as anything else RStudio saves is huge and difficult to transfer. It shows the same behaviour. 

I also transferred the analysis code to R-script.   

Thank you for all your advice!

Best Regards,

Virva

[Chris Watson]

Sorry for the delay. I have discovered the source of the bug and hope to get a new release out by the weekend.

Chris

To view this discussion on the web visit https://groups.google.com/d/msgid/brainGraph-help/16085bbc-648a-43a6-8273-9d457667111dn%40googlegroups.com.

[virva.sa...@gmail.com]

Thank you! Looking forward to the new version!

[virva.sa...@gmail.com]

Hi Chris,

It looks to me that the brainGraph package is not upgraded. I tryed to install a new version using the devtools.  

devtools::install_github('cwatson/brainGraph')

I get an output:

Skipping install of 'brainGraph' from a github remote, the SHA1 (b78bae97) has not changed since last install

I’m I trying to install it from the wrong place? 

Sorry for the extra trouble!

Best,

Virva

[Chris Watson]

I have not had a chance to update yet. I should have some time over the next few days to get around to it.

Chris

To view this discussion on the web visit https://groups.google.com/d/msgid/brainGraph-help/85b51dc5-f39d-4699-94c1-a1404dfe6339n%40googlegroups.com.

[Chris Watson]

Sorry for the delay. I have just pushed the fix to GitHub so you can update the package and it should work properly now.

Chris