DESeq2: Contrasts in multifactor design with interaction terms

xrlg · 07-23-2019, 01:26 PM

Hello,

Please note that I have posted this question on another website, but I did not receive any advice. I would really appreciate some help, and apologize if this is against the rules. I can delete my post if needed.

I am using DESeq2 and I am a beginner with using model matrices and contrasts. I have a design with three variables: tissue, time and phenotype, where the tissue has three levels (X, Y and Z, where X is the reference level), time as three levels (12h, 24h and 48h, where 12h is the reference level), and phenotype has three levels (ctrl, A and B, where ctrl is the reference level). I have added the dput of the design table below my post (first column is just the sample ID). I am using the following formula to consider all possible variables and interaction terms in the dds object:

~ tissue * time * phenotype

The reason I am doing this is because I am developing a function that uses DESeq2 with a special model matrix and returns a dds object, and my goal is to generate a single dds object, and then be able to use the contrasts to retrieve differential gene expression for any combination of these variables. For example, I could then want to get differential gene expression for (tissue Y time 48h phenotype B) vs (tissue Y time 48h phenotype ctrl), or another example could be (tissue X time 24h phenotype A) vs (tissue X time 24h phenotype ctrl). Please note that I DO know how to run DESeq2 for simple pairwise comparisons, but this is not the goal of this post. The question is not about figuring out the appropriate biological question, design, or formula. This is about designing a function that is able to handle a request like the two examples I mentioned.

So right now with this formula the resultsNames(dds) are:

tissueY
tissueZ
time24h
time48h
phenotypeA
phenotypeB
tissueY:time24h
tissueZ:time24h
tissueY:time48h
tissueZ:time48h
tissueY:phenotypeA
tissueZ:phenotypeA
tissueY:phenotypeB
tissueZ:phenotypeB
time24h:phenotypeA
time48h:phenotypeA
time24h:phenotypeB
time48h:phenotypeB
tissueY:time24h:phenotypeA
tissueZ:time24h:phenotypeA
tissueY:time48h:phenotypeA
tissueZ:time48h:phenotypeA
tissueY:time24h:phenotypeB
tissueZ:time24h:phenotypeB
tissueY:time48h:phenotypeB
tissueZ:time48h:phenotypeB

To extract differential expression between, for example, (tissue Y time 48h phenotype B) and (tissue Y time 48h phenotype ctrl), what would be the correct contrasts? And what would it become if were are interested in one of the reference levels, for example differential expression between (tissue X time 24h phenotype A) and (tissue X time 24h phenotype ctrl), where tissue X was set as the reference?

Thank you for your help!

The design deput:

Code:

dput(design) structure(list(tissue = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L), .Label = c("X", "Y", "Z"), class = "factor"), time = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L), .Label = c("12h", "24h", "48h"), class = "factor"), phenotype = structure(c(3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L ), .Label = c("A", "B", "ctrl"), class = "factor")), class = "data.frame", row.names = c("s1", "s2", "s3", "s4", "s5", "s6", "s7", "s8", "s9", "s10", "s11", "s12", "s13", "s14", "s15", "s16", "s17", "s18", "s19", "s20", "s21", "s22", "s23", "s24", "s25", "s26", "s27", "s28", "s29", "s30", "s31", "s32", "s33", "s34", "s35", "s36", "s37", "s38", "s39", "s40", "s41", "s42", "s43", "s44", "s45", "s46", "s47", "s48", "s49", "s50", "s51", "s52", "s53", "s54", "s55", "s56", "s57", "s58", "s59", "s60", "s61", "s62", "s63", "s64", "s65", "s66", "s67", "s68", "s69", "s70", "s71", "s72", "s73", "s74", "s75", "s76", "s77", "s78", "s79", "s80", "s81"))

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Limma multifactor experiment design matrix	sage12	Bioinformatics	0	10-08-2018 12:09 AM
Limma multifactor experiment design matrix	Lugalbanda	Bioinformatics	3	10-19-2016 01:12 AM
Deseq2 interaction terms	denalitastic	Bioinformatics	0	06-07-2016 04:49 PM
Deseq2 multifactor analysis	genomica	Bioinformatics	6	09-26-2015 02:03 AM
DESeq2 model design + contrasts	gamb	Bioinformatics	39	06-25-2015 03:06 PM

07-23-2019, 01:26 PM	#1
xrlg Junior Member Location: Canada Join Date: Jul 2019 Posts: 1	DESeq2: Contrasts in multifactor design with interaction terms Hello, Please note that I have posted this question on another website, but I did not receive any advice. I would really appreciate some help, and apologize if this is against the rules. I can delete my post if needed. I am using DESeq2 and I am a beginner with using model matrices and contrasts. I have a design with three variables: tissue, time and phenotype, where the tissue has three levels (X, Y and Z, where X is the reference level), time as three levels (12h, 24h and 48h, where 12h is the reference level), and phenotype has three levels (ctrl, A and B, where ctrl is the reference level). I have added the dput of the design table below my post (first column is just the sample ID). I am using the following formula to consider all possible variables and interaction terms in the dds object: ~ tissue * time * phenotype The reason I am doing this is because I am developing a function that uses DESeq2 with a special model matrix and returns a dds object, and my goal is to generate a single dds object, and then be able to use the contrasts to retrieve differential gene expression for any combination of these variables. For example, I could then want to get differential gene expression for (tissue Y time 48h phenotype B) vs (tissue Y time 48h phenotype ctrl), or another example could be (tissue X time 24h phenotype A) vs (tissue X time 24h phenotype ctrl). Please note that I DO know how to run DESeq2 for simple pairwise comparisons, but this is not the goal of this post. The question is not about figuring out the appropriate biological question, design, or formula. This is about designing a function that is able to handle a request like the two examples I mentioned. So right now with this formula the resultsNames(dds) are: tissueY tissueZ time24h time48h phenotypeA phenotypeB tissueY:time24h tissueZ:time24h tissueY:time48h tissueZ:time48h tissueY:phenotypeA tissueZ:phenotypeA tissueY:phenotypeB tissueZ:phenotypeB time24h:phenotypeA time48h:phenotypeA time24h:phenotypeB time48h:phenotypeB tissueY:time24h:phenotypeA tissueZ:time24h:phenotypeA tissueY:time48h:phenotypeA tissueZ:time48h:phenotypeA tissueY:time24h:phenotypeB tissueZ:time24h:phenotypeB tissueY:time48h:phenotypeB tissueZ:time48h:phenotypeB To extract differential expression between, for example, (tissue Y time 48h phenotype B) and (tissue Y time 48h phenotype ctrl), what would be the correct contrasts? And what would it become if were are interested in one of the reference levels, for example differential expression between (tissue X time 24h phenotype A) and (tissue X time 24h phenotype ctrl), where tissue X was set as the reference? Thank you for your help! The design deput: Code: dput(design) structure(list(tissue = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L), .Label = c("X", "Y", "Z"), class = "factor"), time = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L), .Label = c("12h", "24h", "48h"), class = "factor"), phenotype = structure(c(3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L, 3L, 1L, 1L, 1L, 2L, 2L, 2L ), .Label = c("A", "B", "ctrl"), class = "factor")), class = "data.frame", row.names = c("s1", "s2", "s3", "s4", "s5", "s6", "s7", "s8", "s9", "s10", "s11", "s12", "s13", "s14", "s15", "s16", "s17", "s18", "s19", "s20", "s21", "s22", "s23", "s24", "s25", "s26", "s27", "s28", "s29", "s30", "s31", "s32", "s33", "s34", "s35", "s36", "s37", "s38", "s39", "s40", "s41", "s42", "s43", "s44", "s45", "s46", "s47", "s48", "s49", "s50", "s51", "s52", "s53", "s54", "s55", "s56", "s57", "s58", "s59", "s60", "s61", "s62", "s63", "s64", "s65", "s66", "s67", "s68", "s69", "s70", "s71", "s72", "s73", "s74", "s75", "s76", "s77", "s78", "s79", "s80", "s81"))