[BioC] Questions about DecideTest and Several groups

[James W. MacDonald]

Hi Elodie,

On 9/11/2012 4:16 AM, Elodie Chapeaublanc wrote:
> Hi,
>
> I have got question about several comparisons and results :
> I have got 4 groups and I want to make several comparisons between my
> groups (gp1 vs gp2, gp1 vs gp3, ...). I tested decideTest in order to
> have easy resume output but I don't understand why in a decideTest I
> obtain results not coording topTable output.

The simple answer is because they are two different functions with 
different outputs (and given that you are using the 'nestedF' argument 
for decideTests(), there is a different assessment of significance as 
well). Have you looked at the help pages for these functions? That 
should clear up any confusion.

For now, note that topTable() outputs the Table of Top genes for a given 
contrast (or alternatively, the top genes by F-statistic if you don't 
specify a coefficient). In contrast, decideTests() outputs a matrix of 
1,0,-1 that indicate significance and direction for each contrast, for 
each gene. This can be used for many things, but the most common I 
suppose is to create Venn diagrams.

Best,

Jim


>
>
> fit<- lmFit(data_i,mat_group)
>
> comparaison<-
> makeContrasts(N-Ta,N-T1,N-T2,Ta-T1,Ta-T2,T1-T2,levels=mat_group)
> fit2<- contrasts.fit(fit,comparaison)
> fit2<- eBayes(fit2)
> top_table<- topTable(fit2,coef=1,number=Inf,adjust="BH")
> rownames(top_table)<- top_table[,1]
>
>
>
>
> ID 	logFC 	AveExpr 	t 	P.Value 	adj.P.Val 	B
> IGF1R 	0.111104650376149 	6.73735754086491 	0.58031824032662
> 0.563347919004554 	0.89178337590619 	-5.80922017018552
> IGF2R 	-0.418361001891785 	7.30781700757716 	-1.89591269242418
> 0.0616198608368053 	0.566570014038297 	-4.30516165436281
> INSR 	-0.104189972020608 	5.88324657993507 	-0.705621827997018
> 0.482494171926321 	0.858963516232972 	-5.73329527131923
> IGF1 	0.0164024330700565 	5.32379121221856 	0.0580159806673495
> 0.953882062264167 	0.993280048337664 	-5.96683923009462
> IGF2 	-2.03692409239562 	7.64672458332555 	-2.78687079989285
> 0.00665914505810975 	0.255548599553939 	-2.46861977043762
> INS 	1.65128828436572 	6.65133927223272 	2.04732749311482
> 0.0439423770026206 	0.508760608526091 	-4.03635411033303
> IGFBP1 	0.111163236117028 	3.7608601464086 	0.990350899711699
> 0.325019419965225 	0.79173927409954 	-5.50673419683211
> IGFBP2 	1.4347995251779 	7.54793353816003 	3.06683251663792
> 0.00295982828163497 	0.16458256036792 	-1.77287225505653
> IGFBP3 	-0.875527979281101 	10.4136969590099 	-1.72885747148627
> 0.0877319854057766 	0.615890064783784 	-4.57990448008225
> IGFBP4 	-1.06382888423619 	10.6781825579171 	-3.59228487330552
> 0.000567234252702273 	0.0858624589183034 	-0.332444870299646
> IGFBP5 	0.0461190965231344 	6.75488133496068 	0.100191694176698
> 0.920445389165695 	0.988013212218042 	-5.96367782236715
> IGFBP6 	-0.333902382597201 	7.18472204506767 	-0.793175371995814
> 0.430046033425277 	0.842849113060914 	-5.67158345971083
> IGFBP7 	-2.06010861248481 	9.26037314643127 	-6.8481256400995
> 1.44348094893832e-09 	2.94975331915545e-06 	11.1853303363173
> IGF2BP3 	-0.0148502449143448 	4.54849676079719 	-0.031861021350583
> 0.974663058942624 	0.997732729342142 	-5.96795306155805
>
>
>
>
>    results<-
> decideTests(fit2,method="nestedF",adjust="BH",p.value=0.05,lfc=0)
> results[intersect(liste_genes_interest,rownames(results)),]
>
>    N - Ta N - T1 N - T2 Ta - T1 Ta - T2 T1 - T2
>     IGF1R        0      0      1       1       1       0
>     IGF2R        0     -1      0       0       0       1
>     INSR         0      0      0       0       0       0
>     IGF1         0      0      0       0       0       0
>     IGF2        -1      0      0       0       1       0
>     INS          0      0      0       0       0       0
>     IGFBP1       0      0      0       0       0       0
>     IGFBP2       1      1      1       0       1       1
>     IGFBP3       0      0      0       0       0       0
>     IGFBP4      -1      0      0       1       1       0
>     IGFBP5       0      0      0       0       0       0
>     IGFBP6       0      0      0       0       0       0
>     IGFBP7      -1     -1     -1      -1      -1       0
>     IGF2BP3      0      0      0       0      -1      -1
>
>
>
>
> _______________________________________________
> Bioconductor mailing list
> 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