[BioC] EdgeR: coefficients for GLM for multiple groups

James W. MacDonald jmacdon at uw.edu
Thu Mar 21 22:17:19 CET 2013


I think you are fundamentally misunderstanding what you are doing. In 
the context of a glm, the way you test for the significance of a 
coefficient is by doing a likelihood ratio test, comparing a model that 
contains the coefficient to a reduced model that does not. So there 
won't be any p-values without making a comparison.

In other words, the glmFit step sets up the model, and the glmLRT step 
does the test for whatever coefficient you care about. By default it 
tests all coefficients. From ?glmLRT:

     glmLRT conducts likelihood ratio tests for one or more
      coefficients in the linear model. If coef is used, the null
      hypothesis is that all the coefficients indicated by coef are
      equal to zero. If contrast is non-null, then the null hypothesis
      is that the specified contrast of the coefficients is equal to
      zero. For example, a contrast of c(0,1,-1), assuming there are
      three coefficients, would test the hypothesis that the second and
      third coefficients are equal.


Best,

Jim



On 3/21/2013 4:55 PM, capricy gao wrote:
> I wonder if there are different versions of userguide/reference manual.
>
> I still couldn't find what I could do to get useful output, like the 
> pvalue for all the coefficients.
>
> The only pvalue I could see is in lrt output...
>
>
> ------------------------------------------------------------------------
> *From:* James W. MacDonald <jmacdon at uw.edu>
> *To:* capricy gao <capricyg at yahoo.com>
> *Cc:* "bioconductor at r-project.org" <bioconductor at r-project.org>
> *Sent:* Thursday, March 21, 2013 3:27 PM
> *Subject:* Re: [BioC] EdgeR: coefficients for GLM for multiple groups
>
> Hi Capricy,
>
> I'm not convinced you have read the user's guide carefully. There is, 
> for example, a section called 'Quick start' that is only seven pages 
> in, that shows the canonical glm analysis steps. Then there are two 
> more worked examples starting on page 52 that show pretty much all the 
> steps required to do a glm analysis.
>
> However, in the code you show below you haven't followed the examples, 
> nor have you used the accessor function that is intended to produce 
> useful output.
>
> Best,
>
> Jim
>
>
>
> On 3/21/2013 3:57 PM, capricy gao wrote:
> > I think I have carefully read both the userguide and reference 
> manual, but couldn't figure it out. Could anybody help me out what 
> arguments should be included? Thanks a lot.
> >
> > In limma fit results: there is something like this:
> > ===============
> > $p.value
> >            Grp1      Grp2vs1
> > Gene 1 0.8604469 6.019156e-05
> > Gene 2 0.2174605 1.673262e-05
> > Gene 3 0.3571957 5.758422e-02
> > Gene 4 0.6789641 1.758690e-01
> > Gene 5 0.4589329 8.223679e-01
> > 95 more rows ...
> > ==============
> >
> > But in edgeR I couldn't find it as I posted previously....
> >
> >
> > ------------------------------------------------------------------------
> > *From:* James W. MacDonald <jmacdon at uw.edu <mailto:jmacdon at uw.edu>>
> > *To:* capricy gao <capricyg at yahoo.com <mailto:capricyg at yahoo.com>>
> > *Cc:* "bioconductor at r-project.org 
> <mailto:bioconductor at r-project.org>" <bioconductor at r-project.org 
> <mailto:bioconductor at r-project.org>>
> > *Sent:* Thursday, March 21, 2013 1:45 PM
> > *Subject:* Re: [BioC] EdgeR: coefficients for GLM for multiple groups
> >
> > Hi Capricy,
> >
> > Just like limma, edgeR has a very comprehensive user's guide that you
> > can consult.
> >
> > library(edgeR)
> > edgeRUsersGuide()
> >
> > Best,
> >
> > Jim
> >
> >
> >
> > On 3/21/2013 2:41 PM, capricy gao wrote:
> > > I am looking at my fit results from GLM of edgeR for my data:
> > >
> > > and assume that $coefficients are related to the expression 
> levels. Just wondering how I can get the p values for those 
> coefficients since I remember in limma, each coefficient will be 
> accompanied by the corresponding t value and p value.
> > >
> > >
> > > Thanks a lot for the help.
> > >
> > > capricy
> > >
> > > ===================
> > >
> > >> design=model.matrix(~0+regroup)
> > >> y<- estimateGLMCommonDisp(y,design)
> > >> y<- estimateGLMTagwiseDisp(y,design)
> > >> fit<- glmFit(y,design)
> > >> fit
> > > An object of class "DGEGLM"
> > > $coefficients
> > >            regroupFR  regroupFA  regroupFM  regroupFP regroupFW  
> regroupMA
> > > GS_14929  -9.543565  -9.267612  -9.310823  -9.167299 -10.62094  
> -9.260763
> > > GS_09776 -10.304430 -10.985146 -10.644154 -10.640469 -10.59615 
> -10.798889
> > > GS_18434 -11.327664 -11.786421 -11.643292 -11.902728 -12.50470 
> -11.654008
> > > GS_08334 -10.789181 -10.271089 -10.511480 -10.375642 -10.39865 
> -10.836647
> > > GS_09550 -10.564167 -10.152571 -10.410428 -10.098825 -10.36302  
> -9.722892
> > >            regroupMM  regroupMP  regroupMW
> > > GS_14929  -8.516256  -8.222225  -9.495394
> > > GS_09776 -10.337434 -10.284703 -10.588544
> > > GS_18434 -11.755905 -11.662276 -11.583950
> > > GS_08334 -10.797529 -10.643409 -10.783104
> > > GS_09550 -10.650495 -10.766462  -9.880359
> > > 15457 more rows ...
> > >
> > > $fitted.values
> > >                  MP        FA      FR        FW        MA      
> FW.1        MM
> > > GS_14929 596.53153 205.75696 84.61834 31.844025 169.27178  
> 82.62258 513.34590
> > > GS_09776  75.81638  36.91137 39.53083 32.643715  36.33823  
> 84.69746  83.04882
> > > GS_18434  19.09812  16.54813 14.19827  4.825192  15.43876  
> 12.51946  20.07922
> > > GS_08334  52.95485  75.41272 24.33895 39.775089  34.99088 
> 103.20054  52.41066
> > > GS_09550  46.82022  84.90529 30.48442 41.218519 106.62251 
> 106.94567  60.71721
> > >                  FP      MA.1        MW        FM      MM.1      
> MW.1      MP.1
> > > GS_14929 78.226172 123.52247 193.95899 122.61523 173.571103 
> 194.05252 568.23926
> > > GS_09776 17.921685  26.51705  64.98434  32.30797  28.080239  
> 65.01567  72.22056
> > > GS_18434  5.064548  11.26610  23.99487  11.88420  6.789129  
> 24.00644  18.19234
> > > GS_08334 23.358730  25.53385  53.48880  36.89427  17.720948  
> 53.51460  50.44331
> > > GS_09550 30.811447  77.80550 131.97315  40.81919  20.529537 
> 132.03679  44.59963
> > >                FM.1      FP.1      FA.1      FR.1
> > > GS_14929 259.44483 324.86875 256.36216 198.47462
> > > GS_09776  68.36129  74.42772  45.98959  92.72063
> > > GS_18434  25.14609  21.03277  20.61808  33.30242
> > > GS_08334  78.06556  97.00745  93.96021  57.08766
> > > GS_09550  86.37041 127.95815 105.78744  71.50204
> > > 15457 more rows ...
> > >
> > > $counts
> > >            MP  FA  FR FW  MA FW.1  MM  FP MA.1  MW  FM MM.1 MW.1 
> MP.1 FM.1 FP.1
> > > GS_14929 221 284 170 23 267  105 209 106  52 218 158  277  170  
> 926  185  211
> > > GS_09776  75  32  17 28  41  96  86  15  23  65  23  27  65  73  
> 87  85
> > > GS_18434  36  22  19  8  21    6  22  7    7  27  13    6  21    
> 2  23  15
> > > GS_08334  44  77  23 50  41  78  78  14  21  61  29    8  46  59  
> 94  132
> > > GS_09550  82  92  45 54 105  75  95  18  79 153  41    8  111  11  
> 86  178
> > >          FA.1 FR.1
> > > GS_14929  159    0
> > > GS_09776  52  143
> > > GS_18434  14  23
> > > GS_08334  92  60
> > > GS_09550  97  39
> > > 15457 more rows ...
> > >
> > > $deviance
> > >  GS_14929  GS_09776  GS_18434  GS_08334  GS_09550
> > > 19.059229  3.629342  9.509356  4.268291  8.809585
> > > 15457 more elements ...
> > >
> > > $df.residual
> > > [1] 9 9 9 9 9
> > > 15457 more elements ...
> > >
> > > $abundance
> > > [1]  -9.081235 -10.552328 -11.717230 -10.579804 -10.233947
> > > 15457 more elements ...
> > >
> > > $design
> > >    regroupFR regroupFA regroupFM regroupFP regroupFW regroupMA 
> regroupMM
> > > 1        0        0        0        0        0        0        0
> > > 2        0        1        0        0        0        0        0
> > > 3        1        0        0        0        0        0        0
> > > 4        0        0        0        0        1        0        0
> > > 5        0        0        0        0        0        1        0
> > >    regroupMP regroupMW
> > > 1        1        0
> > > 2        0        0
> > > 3        0        0
> > > 4        0        0
> > > 5        0        0
> > > 13 more rows ...
> > >
> > > $offset
> > >          [,1]    [,2]    [,3]    [,4]    [,5]    [,6]    [,7]    [,8]
> > > [1,] 14.61341 14.59445 13.9819 14.08233 14.39241 15.03576 14.75727 
> 13.52703
> > > [2,] 14.61341 14.59445 13.9819 14.08233 14.39241 15.03576 14.75727 
> 13.52703
> > > [3,] 14.61341 14.59445 13.9819 14.08233 14.39241 15.03576 14.75727 
> 13.52703
> > > [4,] 14.61341 14.59445 13.9819 14.08233 14.39241 15.03576 14.75727 
> 13.52703
> > > [5,] 14.61341 14.59445 13.9819 14.08233 14.39241 15.03576 14.75727 
> 13.52703
> > >          [,9]    [,10]    [,11]    [,12]  [,13]    [,14]    [,15]  
>   [,16]
> > > [1,] 14.07733 14.76322 14.12002 13.67291 14.7637 14.56482 14.86951 
> 14.95085
> > > [2,] 14.07733 14.76322 14.12002 13.67291 14.7637 14.56482 14.86951 
> 14.95085
> > > [3,] 14.07733 14.76322 14.12002 13.67291 14.7637 14.56482 14.86951 
> 14.95085
> > > [4,] 14.07733 14.76322 14.12002 13.67291 14.7637 14.56482 14.86951 
> 14.95085
> > > [5,] 14.07733 14.76322 14.12002 13.67291 14.7637 14.56482 14.86951 
> 14.95085
> > >          [,17]    [,18]
> > > [1,] 14.81434 14.83441
> > > [2,] 14.81434 14.83441
> > > [3,] 14.81434 14.83441
> > > [4,] 14.81434 14.83441
> > > [5,] 14.81434 14.83441
> > > 15457 more rows ...
> > >
> > > $dispersion
> > > [1] 0.7464955 0.2734189 0.4034924 0.2832934 0.3819788
> > > 15457 more elements ...
> > >
> > > $method
> > > [1] "oneway"
> > >
> > > $samples
> > >    group lib.size norm.factors
> > > MP    MP  2220863            1
> > > FA    FA  2179157            1
> > > FR    FR  1181036            1
> > > FW    FW  1305802            1
> > > MA    MA  1780507            1
> > > 13 more rows ...
> > >    [[alternative HTML version deleted]]
> > >
> > >
> > >
> > > _______________________________________________
> > > Bioconductor mailing list
> > > Bioconductor at r-project.org <mailto:Bioconductor at r-project.org> 
> <mailto:Bioconductor at r-project.org <mailto:Bioconductor at r-project.org>>
> > > https://stat.ethz.ch/mailman/listinfo/bioconductor
> > > Search the archives: 
> http://news.gmane.org/gmane.science.biology.informatics.conductor
> >
> > -- James W. MacDonald, M.S.
> > Biostatistician
> > University of Washington
> > Environmental and Occupational Health Sciences
> > 4225 Roosevelt Way NE, # 100
> > Seattle WA 98105-6099
> >
> >
> >
>
> -- James W. MacDonald, M.S.
> Biostatistician
> University of Washington
> Environmental and Occupational Health Sciences
> 4225 Roosevelt Way NE, # 100
> Seattle WA 98105-6099
>
>
>

-- 
James W. MacDonald, M.S.
Biostatistician
University of Washington
Environmental and Occupational Health Sciences
4225 Roosevelt Way NE, # 100
Seattle WA 98105-6099



More information about the Bioconductor mailing list