Search
Question: DESeq2
0
4.3 years ago by
Guest User12k
Guest User12k 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","breed 1","breed2","breed2","breed2","breed1","breed1","breed1","breed2","bre ed2","breed2","breed1","breed1","breed1","breed2","breed2","breed2"), day = c("D1","D1","D1","D1","D1","D1","D2","D2","D2","D2","D2","D2","D 3","D3","D3","D3","D3","D3","D4","D4","D4","D4","D4","D4")) 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 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? 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.
modified 4.3 years ago by Michael Love18k • written 4.3 years ago by Guest User12k
0
4.3 years ago by
Michael Love18k
United States
Michael Love18k wrote:
On Fri, Apr 4, 2014 at 11:30 AM, aroa [guest] <guest@bioconductor.org>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","bree d2","breed2","breed2","breed1","breed1","breed1","breed2","breed2","br eed2","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")) > âside note: i would recommend putting the phenotypic data in a CSV file and reading it in with read.csv.â > 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 > > â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? > âTo compare different breeds at each time point, you need to use the model with the interaction termâ. 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")) âMikeâ 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]]
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 <michaelisaiahlove@gmail.com>: > > On Fri, Apr 4, 2014 at 11:30 AM, aroa [guest] <guest@bioconductor.org>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","bre ed2","breed2","breed2","breed1","breed1","breed1","breed2","breed2","b reed2","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]]
ADD REPLYlink written 4.3 years ago by AROA SUÁREZ VEGA20
âhi Aroa, With the design "~ breed + day", you assume that the breed effect is the same for each day, and the day effect is the same for each breed. So there is no way to get different breed effects for each time point, or vice versa with this design.â If you investigated and found that the interaction terms were small/not significant, then it means that there are not specific breed effects for each day, but that the main effect is sufficient. Mike On Fri, Apr 4, 2014 at 12:10 PM, AROA SUÃREZ VEGA <asuav@unileon.es> wrote: > 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 SuÃ¡rez Vega > PhD student > Dpto. ProducciÃ³n 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 <michaelisaiahlove@gmail.com>: > >> >> On Fri, Apr 4, 2014 at 11:30 AM, aroa [guest] <guest@bioconductor.org>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","br eed2","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]]

Content
Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.