Interaction using a single factor model DESeq2
1
0
Entering edit mode
slcno • 0
@500b8585
Last seen 16 days ago
France

Hello everyone,

I have a question related to the analysis of those genes that are differentially expressed in three mutants single KO A, single KO B and double KO C where C is a combination of mutant A + mutant B.

I am interested in analyzing the effect of combining mutant A and B to their individual effect.

So, I considered an additive effect if the double mutant is equal to the sum of the two simple mutants (δ = C - A - B = 0) and an interaction effect if δ ≠ 0.

I used the table of counts for all genes and I used DESeq2 to normalize. Then I used contrast to determine δ

This is my code:

data <- read.table("total_counts.txt",header=T,row.names=1)
groups <- factor(rep(c("CTL", "mutantA", "mutantB", "mutantC"), each=3))
sampleInfo <- data.frame(groups,row.names=colnames(data))
dds <- DESeqDataSetFromMatrix(countData = data, colData = sampleInfo, design = ~groups)
dds$groups <- relevel(dds$groups, "CTRL")
colData(dds)

DataFrame with 12 rows and 1 column
                         groups
                        <factor>
siCtrl_1                    CTRL   
siCtrl_2                    CTRL   
siCtrl_3                    CTRL   
mutantA_1               mutantA
mutantA_2               mutantA
...                         ...
mutantB_2               mutantB       
mutantB_3               mutantB       
mutantC_1               mutantC
mutantC_2               mutantC
mutantC_3               mutantC

design(dds) <- ~ 1 + groups
dds <- DESeq(dds)

estimating size factors
estimating dispersions
gene-wise dispersion estimates
mean-dispersion relationship
final dispersion estimates
fitting model and testing

resultsNames(dds)

[1] "Intercept"                     "groups_mutantA_vs_CTRL"        "groups_mutantB_vs_CTRL"        "groups_mutantC_vs_CTRL"

mod_mat <- model.matrix(design(dds), colData(dds))
mod_mat

                 (Intercept)        groupsmutantA           groupsmutantB           groupsmutantC
siCtrl_1                1                       0                       0                       0
siCtrl_2                1                       0                       0                       0
siCtrl_3                1                       0                       0                       0
mutantA_1               1                       1                       0                       0
mutantA_2               1                       1                       0                       0
mutantA_3               1                       1                       0                       0
mutantB_1               1                       0                       1                       0
mutantB_2               1                       0                       1                       0
mutantB_3               1                       0                       1                       0
mutantC_1               1                       0                       0                       1
mutantC_2               1                       0                       0                       1
mutantC_3               1                       0                       0                       1
attr(,"assign")
[1] 0 1 1 1
attr(,"contrasts")
attr(,"contrasts")$groups
[1] "contr.treatment"

A <- colMeans(mod_mat[dds$groups == "mutantA", ])
B <- colMeans(mod_mat[dds$groups == "mutantB", ])
C <- colMeans(mod_mat[dds$groups == "mutantC", ])
CTRL <- colMeans(mod_mat[dds$groups == "CTRL", ])
res <- results(dds, contrast = C - A - B)

res

log2 fold change (MLE): -1,-1,-1,+1 
Wald test p-value: -1,-1,-1,+1 
DataFrame with 58721 rows and 6 columns
                    baseMean log2FoldChange     lfcSE        stat    pvalue      padj
                   <numeric>      <numeric> <numeric>   <numeric> <numeric> <numeric>
ENSG00000000003.14  1235.661      -10.28328 0.0878839   -117.0099         0         0
ENSG00000000005.5      0.000             NA        NA          NA        NA        NA
ENSG00000000419.12  1625.580      -10.70786 0.0909223   -117.7694         0         0
ENSG00000000457.13   509.194       -8.85948 0.1080919    -81.9625         0         0
ENSG00000000460.16  1054.786       -9.75485 0.0815451   -119.6252         0         0
...                      ...            ...       ...         ...       ...       ...
ENSG00000285990.1    1.28962       -1.13755 1.8312013   -0.621204  0.534465  0.628871
ENSG00000285991.1  692.47452       -9.63571 0.0863505 -111.588309  0.000000  0.000000
ENSG00000285992.1    0.00000             NA        NA          NA        NA        NA
ENSG00000285993.1    0.00000             NA        NA          NA        NA        NA
ENSG00000285994.1    2.82204       -1.18911 1.4336056   -0.829452  0.406849  0.502778

sessionInfo( )

What I don't understand is the relationship between the log2FC and the δ. How can I determine δ from the log2FC? If I want to determine only genes with an interaction effect, or an additive effects, should I use the total table of count for all genes or should I first remove all genes where there is not significative difference between mutant A vs mutant C / mutant B vs mutant C / CTRL vs mutant C?

Thanks in advance!

Sophie

synergy DESeq2 RNA-seq contrast • 126 views
ADD COMMENT
0
Entering edit mode
@mikelove
Last seen 3 hours ago
United States

Unfortunately, I have to save my time to address software related questions. For questions about how to set up your statistical analysis, I recommend consulting with a local statistician or someone familiar with running linear models in R (DESeq2 uses the same linear model setup as base R).

ADD COMMENT

Login before adding your answer.

Traffic: 507 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

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

Powered by the version 2.3.6