GraphVar group1 vs. group2

Submitted by iidaka on

Thank you very much for providing such useful software.

I downloaded the latest version of GraphVar and am using it for analyzing small world network. My question is about between group analysis of Statistics module. When analyzing the difefrence in raw matrix between group1 and group2, results viewer may show  an inverse pattern of the difefrence, it means that difference matrix of group1 < group2 looks like that of group1 > group2. In addition, I extracted numerical data from smallworldness_bu_xxx.mat and plotted the mean values as a function of threshold (from 1.0 to 5.0) , but the plot of group1 and group2 was opposite of that shown in results viewer. Your help is very much appreciated.

tetsuya

 

Johann Kruschwitz

Fri, 11/07/2014 - 11:42

Dear Tetsuya,

this seems strange and should not happen. I tried to replicate what you describe but I get the same results as in the results viewer when I extract the data. To make sure we really talk about the same thing this is what I did with the sample data from the sample workspace:

1. load all 30 subjects

2. calculate a group difference between male and female in the raw matrix on the connectivity between Precentral_L and Precentral_R

3. the contrast Group(female)<Group(male) should result in a difference of 0.044 (which means that males have a higher connectivity)

4. now extract the respective cells in the sample connectivity matrices for Precentral_L and R and plot these according to the subject in e.g. in the variable sheet

5. If I now manually calculate the group difference in connectivity from the extracted values (males-females) I get exactly the same difference of 0.044, telling me that males>females (as in the results viewer)

6. I did the same with smallworldness and the weighted clustering coefficient (calculate end export via the "calculate and export" button in the network calculations panel) and also manually run a group comparison. Again, the results mirror the results from the results viewer. At least for the weighted clustering coefficient it is also males>females with a positive difference of 0.02 at threshold .048 (as smallworldness is claculated from random networks our results could differ, but for me it is also males>females with a difference of 0.02).

Please try this out to see if you get the same results as I do - if you get the same results, results from your analysis (with other data) should also be correct.

Let me know what comes out!

Johann

 

Dear Johann
 
Thank you for response.
 
Yes, you are exactly correct.
I changed name of the groups in the Variable.csv file from 1/2 to pre/post, this works fine.
 
tetsuya
 
 
 
Sent: Friday, November 07, 2014 8:42 PM
Subject: Re: [RFMRI] GraphVar group1 vs. group2
 
[To post a comment, please reply to rfmri.org@gmail.com ABOVE this line]
 
Commented by Johann Kruschwitz (Johann Kruschwitz)

Dear Tetsuya,

this seems strange and should not happen. I tried to replicate what you describe but I get the same results as in the results viewer when I extract the data. To make sure we really talk about the same thing this is what I did with the sample data from the sample workspace:

1. load all 30 subjects

2. calculate a group difference between male and female in the raw matrix on the connectivity between Precentral_L and Precentral_R

3. the contrast Group(female)<Group(male) should result in a difference of 0.044 (which means that males have a higher connectivity)

4. now extract the respective cells in the sample connectivity matrices for Precentral_L and R and plot these according to the subject in e.g. in the variable sheet

5. If I now manually calculate the group difference in connectivity from the extracted values (males-females) I get exactly the same difference of 0.044, telling me that males>females (as in the results viewer)

6. I did the same with smallworldness and the weighted clustering coefficient (calculate end export via the "calculate and export" button in the network calculations panel) and also manually run a group comparison. Again, the results mirror the results from the results viewer. At least for the weighted clustering coefficient it is also males>females with a positive difference of 0.02 at threshold .048 (as smallworldness is claculated from random networks our results could differ, but for me it is also males>females with a difference of 0.02).

Please try this out to see if you get the same results as I do - if you get the same results, results from your analysis (with other data) should also be correct.

Let me know what comes out!

Johann

 


Online version of this post: http://rfmri.org/comment/3406#comment-3406


 

 

Best wishesDear Johann,

 

Hi

I am trying to analysis my data including 35 subjects with group comparison, I used the mutual information to create the graphs but I have serious problems with following sample settings. I was wondering if you could help me.

Best wishes

Index exceeds matrix dimensions.

 

Error in executeStats (line 83)

                            if isnumeric(VarData{1})

 

Error in GraphVar_corr (line 89)

    [Files,N] =

    executeStats(allTasks,dialogData.statsType,dialogData.functionList,dialogData.thresholds,dialogData.VarList,dialogData.CovarList,nRandom,dialogData.nShuffel,0,dialogData.ConnectivityT

    Error in GraphVar>Btn_calcCorr_Callback (line 73)

    GraphVar_corr(handles,allTask)

 

Error in gui_mainfcn (line 96)

        feval(varargin{:});

 

Error in GraphVar (line 40)

    gui_mainfcn(gui_State, varargin{:});

 

Error in

@(hObject,eventdata)GraphVar('Btn_calcCorr_Callback',hObject,eventdata,guidata(hObject))

 

 

Error while evaluating uicontrol Callback

 

Warning: Possible deprecated use of set on a Java object with an

HG Property 'MouseMovedCallback'. 

> In GraphVar>list_brainvars_var_Callback at 63

  In gui_mainfcn at 96

  In GraphVar at 40

  In @(hObject,eventdata)GraphVar('list_brainvars_var_Callback',hObject,eventdata,guidata(hObject)) 

Index exceeds matrix dimensions.

 

Error in executeStats (line 83)

                            if isnumeric(VarData{1})

 

Error in GraphVar_corr (line 89)

    [Files,N] =

    executeStats(allTasks,dialogData.statsType,dialogData.functionList,dialogData.thresholds,dialogData.VarList,dialogData.CovarList,nRandom,dialogData.nShuffel,0,dialogData.ConnectivityT

    Error in GraphVar>Btn_calcCorr_Callback (line 73)

    GraphVar_corr(handles,allTask)

 

Error in gui_mainfcn (line 96)

        feval(varargin{:});

 

Error in GraphVar (line 40)

    gui_mainfcn(gui_State, varargin{:});

 

Error in

@(hObject,eventdata)GraphVar('Btn_calcCorr_Callback',hObject,eventdata,guidata(hObject))

 

 

Error while evaluating uicontrol Callback

 

>

Johann Kruschwitz

Fri, 06/12/2015 - 17:33

Hey, frim this message i dont exactly know whats going on. One of my guesses is that you did not select the subject identifier 'subj_id' as displayed in your variable list as identifier. At least, if this is your identifier, it should not be displayed in the list. If you load your variable sheet, select the left row button for spwcification of your id. Otherwise you would have to send me one of the MI matrices and your variable sheet and i could have a look whats going on. best, Johann

Greetings,

I am receiving a similar error as was previously posted within this thread:

Index exceeds matrix dimensions.

Error in executeStats (line 83).... etc.

Error in GraphVar corr (line 89)

I tried what was suggested regarding the specification of the identifier but was not able to resolve the problem.

 

 

Greetings,

I am receiving a similar error as was previously posted within this thread:

Index exceeds matrix dimensions.

Error in executeStats (line 83).... etc.

Error in GraphVar corr (line 89)

I tried what was suggested regarding the specification of the identifier but was not able to resolve the problem.

 

 

Forums