chooseTopHubInEachModule

Question

[wes]

I'm following the Tutorials for the WGCNA package (Tutorial I) for network construction. A total of 18 modules were generated. Next, I would like to identify top hub gene in each module.

If I followed the previous post, more that 18 hub gene were identified which is not really tally with the previous finding of 18 modules generated.

colorh = labels2colors(rownames(datExpr))

 chooseTopHubInEachModule(
    datExpr, 
    colorh, 
    omitColors = "grey", 
    power = 2, 
    type = "signed", 
    )


 antiquewhite1   antiquewhite2   antiquewhite4         bisque4           black 
  "MMT00025271"   "MMT00030705"   "MMT00067823"   "MMT00057439"   "MMT00081213" 
           blue           blue2           blue3           blue4      blueviolet 
  "MMT00028861"   "MMT00054781"   "MMT00004171"   "MMT00046684"   "MMT00071699" 
          brown          brown1          brown2          brown4      chocolate4 
  "MMT00053547"   "MMT00044818"   "MMT00042999"   "MMT00078844"   "MMT00028559" 
          coral          coral1          coral2          coral3          coral4 
  "MMT00036790"   "MMT00005236"   "MMT00020591"   "MMT00052548"   "MMT00072237" 
           cyan       darkgreen        darkgrey     darkmagenta  darkolivegreen 
  "MMT00056768"   "MMT00016342"   "MMT00004625"   "MMT00067294"   "MMT00048191" 
darkolivegreen1 darkolivegreen2 darkolivegreen4      darkorange     darkorange2 
  "MMT00059782"   "MMT00071677"   "MMT00064617"   "MMT00054037"   "MMT00076332" 
        darkred   darkseagreen2   darkseagreen3   darkseagreen4   darkslateblue 
  "MMT00056866"   "MMT00012463"   "MMT00052488"   "MMT00082420"   "MMT00057450" 
  darkturquoise      darkviolet        deeppink      firebrick2      firebrick3 
  "MMT00047884"   "MMT00058337"   "MMT00006397"   "MMT00075508"   "MMT00074641" 
     firebrick4     floralwhite           green          green4     greenyellow 
  "MMT00033297"   "MMT00023471"   "MMT00022665"   "MMT00053218"   "MMT00001154" 
         grey60        honeydew       honeydew1      indianred2      indianred3 
  "MMT00069884"   "MMT00011559"   "MMT00071941"   "MMT00078034"   "MMT00077345" 
     indianred4           ivory  lavenderblush1  lavenderblush2  lavenderblush3 
  "MMT00024107"   "MMT00046878"   "MMT00080624"   "MMT00058428"   "MMT00055467" 
     lightblue3      lightblue4      lightcoral       lightcyan      lightcyan1 
  "MMT00066256"   "MMT00040558"   "MMT00036444"   "MMT00007435"   "MMT00003410" 
     lightgreen      lightpink2      lightpink3      lightpink4   lightskyblue4 
  "MMT00015397"   "MMT00072076"   "MMT00020996"   "MMT00023860"   "MMT00004034" 
 lightslateblue  lightsteelblue lightsteelblue1     lightyellow         magenta 
  "MMT00034010"   "MMT00017441"   "MMT00045533"   "MMT00063363"   "MMT00053716" 
       magenta3        magenta4          maroon    mediumorchid    mediumpurple 
  "MMT00015549"   "MMT00017876"   "MMT00048511"   "MMT00036275"   "MMT00041355" 
  mediumpurple1   mediumpurple2   mediumpurple3   mediumpurple4    midnightblue 
  "MMT00011887"   "MMT00076735"   "MMT00060741"   "MMT00036913"   "MMT00056798" 
      mistyrose     navajowhite    navajowhite1    navajowhite2          orange 
  "MMT00044489"   "MMT00015548"   "MMT00043149"   "MMT00070510"   "MMT00075941" 
      orangered      orangered1      orangered3      orangered4   paleturquoise 
  "MMT00031650"   "MMT00061815"   "MMT00082585"   "MMT00032698"   "MMT00026361" 
 palevioletred1  palevioletred2  palevioletred3            pink           pink3 
  "MMT00021805"   "MMT00037437"   "MMT00005189"   "MMT00016084"   "MMT00080789" 
          pink4            plum           plum1           plum2           plum3 
  "MMT00003071"   "MMT00067910"   "MMT00048205"   "MMT00042535"   "MMT00033419" 
          plum4          purple             red       royalblue     saddlebrown 
  "MMT00080097"   "MMT00072785"   "MMT00035101"   "MMT00038868"   "MMT00010542" 
         salmon         salmon1         salmon2         salmon4         sienna2 
  "MMT00001185"   "MMT00080541"   "MMT00071245"   "MMT00051771"   "MMT00050328" 
        sienna3         sienna4         skyblue        skyblue1        skyblue2 
  "MMT00051480"   "MMT00049308"   "MMT00010513"   "MMT00080515"   "MMT00079155" 
       skyblue3        skyblue4       slateblue       steelblue             tan 
  "MMT00007859"   "MMT00027667"   "MMT00056001"   "MMT00042156"   "MMT00034198" 
           tan4         thistle        thistle1        thistle2        thistle3 
  "MMT00034355"   "MMT00024286"   "MMT00042538"   "MMT00002004"   "MMT00036605" 
       thistle4       turquoise          violet           white          yellow 
  "MMT00049901"   "MMT00031883"   "MMT00026397"   "MMT00057086"   "MMT00025736" 
        yellow2         yellow3         yellow4     yellowgreen 
  "MMT00028642"   "MMT00003058"   "MMT00039749"   "MMT00043936" 
`````````````````````

Hence, I would like to know if the correct way of identify hub gene should be as below?

moduleLabels = net$colors
moduleColors = labels2colors(net$colors)

chooseTopHubInEachModule(
+          datExpr, 
+          moduleColors, 
+          omitColors = "grey", 
+          power = 6, 
+          type = "signed")

        black          blue         brown          cyan         green   greenyellow 
"MMT00026784" "MMT00037643" "MMT00006713" "MMT00076726" "MMT00005744" "MMT00052488" 
       grey60     lightcyan    lightgreen       magenta  midnightblue          pink 
"MMT00052525" "MMT00058688" "MMT00059368" "MMT00051830" "MMT00007970" "MMT00048286" 
       purple           red        salmon           tan     turquoise        yellow 
"MMT00020330" "MMT00052927" "MMT00060232" "MMT00010513" "MMT00061093" "MMT00042783" 


``````````````````````````

[andres.firrincieli]

The second chunck of code is correct. You get one one hub for each module you have in moduleColors.

How many modules do you have?

[wes]

I obtained 18 modules (excluding the grey modules) from the network construction and hence 2nd code should be correct. Just that I could not understand why the previous post suggest

colorh = labels2colors(rownames(datExpr))

On the other hand, I also converted the WGCNA object to Cytoscape and use NetworkAnalyzer plugin to identify hub gene (gene with highest degree) to cross-check. Some of the hub gene identified from Rscript (chooseTopHubInEachModule) matched with the output of the NetworkAnalyzer (highest degree that ranked no.1) while some hub gene that identified in Rscript ranked top10 in NetworkAnalyzer instead of ranking no.1. Just wondering what cause this slight differences?

[andres.firrincieli]

Just that I could not understand why the previous post suggest

That first chunck of code is just wrong. With colorh = labels2colors(rownames(datExpr)) you get a vector of colors corresponding to each row in datExpr. So, if datExpr has 300 samples (rownames) you will get 300 different colors in colorh. This is not how modules are detected in WGCNA. The label2colors function was implemented because in the WGCNA pipeline modules are first named with numbers. Then, these numbers are converted in to colors with labels2colors

Just wondering what cause this slight differences?

Different ways to calculate the hubiness of a node in a network. In WGCNA the intramodular connectivity of a node is the sum of connection strengths (correlation values) to other nodes within the same module. The gene with the highest intramodular connectivity is the TopHub

[wes]

Thanks for the explanation.

In WGCNA the intramodular connectivity of a node is the sum of connection strengths (correlation values) to other nodes within the same module. May I know what is the formula for the statement above?

intramodularConnectivity(adjMat, colors, scaleByMax = FALSE)

intramodularConnectivity.fromExpr(datExpr, colors, 
              corFnc = "cor", corOptions = "use = 'p'",
              weights = NULL,
              distFnc = "dist", distOptions = "method = 'euclidean'",
              networkType = "unsigned", power = if (networkType=="distance") 1 else 6,
              scaleByMax = FALSE,
              ignoreColors = if (is.numeric(colors)) 0 else "grey",
              getWholeNetworkConnectivity = TRUE)

or

[andres.firrincieli]

That equation is implemented in intramodularConnectivity and intramodularConnectivity.fromExpr to calculate the intramodular connectivity of a gene