Thank you very much for your answer, so if I understand well, even I don't have interaction I need to use the model with the interaction term, isn't it? Aroa Suárez Vega PhD student Dpto. Producción Animal Facultad de Veterinaria Campus de Vegazana s/n 24071 Leon Telef. 987291000 Ext. 5296 2014-04-04 17:59 GMT+02:00 Michael Love : > > On Fri, Apr 4, 2014 at 11:30 AM, aroa [guest] wrote: > >> >> We have and experiment to measure the differences in the milk >> transcriptome for two breeds. We have RNA-seq samples for these two breeds >> in 4 different time points, for each breed and time point we have three >> replicates. >> >> Day1 Day2 Day3 Day4 >> Breed1 3 3 3 3 >> Breed2 3 3 3 3 >> >> >> We want to use DESeq2 to extract the differential expressed (DE) genes >> for each time point between the two breeds and the DE genes for each breed >> comparing the different time points. >> We have tested for the interaction (~breed+day+breed:day) and we cannot >> find interaction between the breeds. >> Now we are running the model like this: >> design=data.frame(row.names = colnames(milk), breed = c("breed1", >> "breed1", >> "breed1","breed2","breed2","breed2","breed1","breed1","breed1","breed2","breed2","breed2","breed1","breed1","breed1","breed2","breed2","breed2","breed1","breed1","breed1","breed2","breed2","breed2"), >> day = >> c("D1","D1","D1","D1","D1","D1","D2","D2","D2","D2","D2","D2","D3","D3","D3","D3","D3","D3","D4","D4","D4","D4","D4","D4")) >> > > > side note: i would recommend putting the phenotypic data in a CSV file > and reading it in with read.csv. > > > > >> dds<- DESeqDataSetFromMatrix(countData= milk, colData= design, design= ~ >> breed + day) >> dds$breed<- factor(dds$breed, levels=c("breed1","breed2")) >> dds$day<- factor(dds$day, levels=c("D1","D2","D3","D4")) >> dds<-DESeq(dds, betaPrior=FALSE) >> resultsNames(dds) >> [1] "Intercept" "breed_breed2_vs_breed1" "day_D2_vs_D1" >> [4] "day_D3_vs_D1" "day_D4_vs_D1" >> >> We would like to know what is the meaning of the resultsNames?, we >> understand them like this: >> Intercept: breed1D1 >> breed_breed2_vs_breed1: breed2-breed1 >> day_D2_vs_D1: (D2-D1)breed1 >> day_D3_vs_D1: (D3-D1)breed1 >> day_D4_vs_D1: (D4-D1)breed1 >> >> > If you fit a model without interactions, then the last three terms here > are, e.g. D2 - D1 across both breeds, not specifically for breed 1. > > > >> And how we can make the contrast to obtain the results that we want, that >> it is comparing the different breeds in each time point and the different >> time points in each breed? >> > > To compare different breeds at each time point, you need to use the model > with the interaction term. Interaction terms are how you encode the > hypothesis, say, that the "breed 2 vs 1 on day 2" effect is not simply the > sum (in log2FC space) of the main "breed 2 vs 1" effect and the main "day 2 > vs day 1" effect. > > In version >= 1.3, we have simplified the extraction of results for > contrasts like this. This version will be released in 10 days, so I'd > prefer to recommend you upgrade to this version. If you need to stick with > v1.2 just email me and I can try to give details. For version 1.3, you > would fit a model with an interaction term. Then the breed 2 vs 1 effect on > day 2 can be done with a list() argument to contrasts (see ?results in > DESeq2 version >= 1.3 for more details): > > results(dds, contrast=list(c("breed_breed2","breedbreed2.dayD2"), > c("breed_breed1","breedbreed1.dayD2")) > > Mike > > > > Thank you in advance. >> >> >> -- output of sessionInfo(): >> >> R version 3.0.3 (2014-03-06) >> Platform: i386-w64-mingw32/i386 (32-bit) >> >> locale: >> [1] LC_COLLATE=French_France.1252 LC_CTYPE=French_France.1252 >> [3] LC_MONETARY=French_France.1252 LC_NUMERIC=C >> [5] LC_TIME=French_France.1252 >> >> attached base packages: >> [1] parallel stats graphics grDevices utils datasets methods >> base >> >> other attached packages: >> [1] gplots_2.12.1 RColorBrewer_1.0-5 DESeq2_1.2.10 >> [4] RcppArmadillo_0.4.100.2.1 Rcpp_0.11.1 >> GenomicRanges_1.14.4 >> [7] XVector_0.2.0 IRanges_1.20.7 >> BiocGenerics_0.8.0 >> [10] BiocInstaller_1.12.0 >> >> loaded via a namespace (and not attached): >> [1] annotate_1.40.1 AnnotationDbi_1.24.0 Biobase_2.22.0 >> bitops_1.0-6 >> [5] caTools_1.16 DBI_0.2-7 gdata_2.13.2 >> genefilter_1.44.0 >> [9] grid_3.0.3 gtools_3.3.1 KernSmooth_2.23-10 >> lattice_0.20-27 >> [13] locfit_1.5-9.1 RSQLite_0.11.4 splines_3.0.3 >> stats4_3.0.3 >> [17] survival_2.37-7 tools_3.0.3 XML_3.98-1.1 >> xtable_1.7-3 >> >> -- >> Sent via the guest posting facility at bioconductor.org. >> >> _______________________________________________ >> Bioconductor mailing list >> Bioconductor@r-project.org >> https://stat.ethz.ch/mailman/listinfo/bioconductor >> Search the archives: >> http://news.gmane.org/gmane.science.biology.informatics.conductor >> > > [[alternative HTML version deleted]]