galeb abu-ali | 28 Jun 15:53 2013
Picon

RNA-seq: interaction effect with DESeq

Hi,

this is my first post on BiocR, I learned a lot by reading existing posts.
I am trying to use DESeq to infer differential gene expression in a
2-factor design of an RNA-seq expt.  Factor 1 (strain) and Factor 2
(condition) each have 4 levels; each level has 3 bioreps totaling to 48
RNA-seq samples.  I got good results from within-factor analysis, and am
now trying to identify an interaction effect of factor1 and factor2 on gene
expression.  This is the code i used:

> fit1 = fitNbinomGLMs( cdsFull, count ~ strain + condition + strain :
condition )
> fit0 = fitNbinomGLMs( cdsFull, count ~ strain + condition )
> pvalsGLM = nbinomGLMTest( fit1, fit0 )
> padjGLM = p.adjust( pvalsGLM, method='BH' )
> DEresults <- transform( fit1, pval=pvalsGLM, padj=padjGLM )
> head( DEresults[ order( DEresults$padj ), ] )

Attached is the output of head.
What I am asking help with is how to determine which of these interaction
comparisons are significant.  Do I need to compare each factor1 level with
each factor2 level, ie pairwise strain : condition comparisons, or is there
a way to extract this from the existing data?  Can this be done in DESeq or
should I be looking at another package?

Your insight is greatly appreciated.

thanks

galeb
	X.Intercept.	strain2	strain3	strain4	condition2	condition3	condition4	strain2.condition2	strain3.condition2	strain4.condition2	strain2.condition3	strain3.condition3	strain4.condition3	strain2.condition4	strain3.condition4	strain4.condition4	deviance	converged	pval	padj
gene_0477	9.3980690498	-1.6791314581	-7.002798797	-1.7106317853	0.1292043759	1.004851704	0.9593594984	0.4460683887	1.3333587254	0.5430856124	1.0467756694	4.3566446059	1.4022027955	1.0247376795	4.0497624618	1.5042551291	66.9551796174	TRUE	0	0
gene_0899	8.8283772691	1.3201257541	3.6966878444	2.0598916209	-0.2863640307	-3.1862934861	-3.6320668852	-0.433159713	0.1183208245	-0.3313504214	-1.3089286975	1.1287940418	-1.8929916603	-0.6393882394	1.6738458015	-1.0571110161	36.5611791438	TRUE	0	0
gene_3193	9.504666246	-0.1170027097	-4.1303583211	-0.6022719781	0.09498906	0.526477533	0.574614423	0.0826986394	0.1735329192	0.2855168364	0.2909434165	2.5168302886	0.6268837601	0.1252186408	2.4662137035	0.5265930419	20.2189071609	TRUE	0	0
gene_3190	9.7134519192	-1.2708138946	-4.6197170035	-1.4720599481	-0.2505335899	0.3404216725	0.5225965378	0.4759426442	0.0465311861	0.536341491	0.833730625	2.3164668939	0.8533719617	0.5648146485	2.0587035777	1.0514146972	32.5110294464	TRUE	2.19E-014	2.22E-011
gene_3192	8.3830050576	-0.6638001747	-4.1436143915	-0.952456834	0.0808601394	0.5629341824	0.4775418862	0.2657938573	-0.1631299939	0.3539415549	0.4834241261	2.2290320256	0.7099500394	0.4530668099	2.3089386737	0.7734138303	29.3662998474	TRUE	8.77E-014	7.13E-011
gene_0031	9.3983124954	0.4919132644	-0.1754119696	0.5397965786	0.3580852136	-2.7711159019	-3.7408103493	0.1876451807	-0.0149511116	-0.1168719638	-0.6959492519	1.3440430005	-0.6036571287	-0.159308005	1.8053857034	-0.4823270854	49.8376020899	TRUE	1.22E-013	8.30E-011
_______________________________________________
Bioconductor mailing list
Bioconductor@...
https://stat.ethz.ch/mailman/listinfo/bioconductor
Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor

Gmane