extracting within time-point differences from time-series model
1
0
Entering edit mode
serpalma.v ▴ 60
@serpalmav-8912
Last seen 13 days ago
Germany

Hello!

thank you for the nice documentation and the DESeq2 package.

I have an experiment with three timepoints (duration) and three treatments (conc_nM). There are also four biological replicates (animal) repeated at each duration-treatment combination.

This is the experiment design:


|         |  0| 100| 250|
|:--------|--:|---:|---:|
|3H       |  4|   4|   4|
|12H      |  4|   4|   4|
|3D       |  4|   4|   4|
|21D      |  4|   4|   4|


I would like to identify differentially expressed genes within each time-point and also to identify genes whose response to treatment is influenced by the time-point:

For this, I applied the following procedure to estimate coefficients:

dds <- DESeqDataSetFromMatrix(countData = gene_cts, colData = sample_info, design = ~ animal + duration + conc_nM + duration:conc_nM)

dds_2 <- DESeq(dds, test = "LRT", reduced = ~animal + duration + conc_nM)


Then I get the following coefficients:

> resultsNames(dds_2)
[1] "Intercept"              "animal_pig_2_vs_pig_1"  "animal_pig_3_vs_pig_1"  "animal_pig_4_vs_pig_1"  "duration_12H_vs_3H"     "duration_3D_vs_3H"
[7] "duration_21D_vs_3H"     "conc_nM_100_vs_0"       "conc_nM_250_vs_0"       "duration12H.conc_nM100" "duration3D.conc_nM100"  "duration21D.conc_nM100"
[13] "duration12H.conc_nM250" "duration3D.conc_nM250"  "duration21D.conc_nM250"


To find differentialy expressed genes within timepoints (i.e. 12h), I could do this:

results(dds_2, name = "duration12H.conc_nM100")

results(dds_2, name = "duration12H.conc_nM250")

results(dds_2, contrast = list("duration12H.conc_nM250","duration12H.conc_nM100"))


But I am not sure how to get the results for time-point 3h

My questions are:

• Is the data modelled correctly?

• How to get coefficients for time-point 3h? Is this implicit in the coefficients conc_nM_100_vs_0 and conc_nM_250_vs_0?

• How to extract the p-values for genes whose response to treatment is influenced by time?

Thanks!

DESeq2 TimeCourse • 70 views
0
Entering edit mode
@mikelove
Last seen 2 hours ago
United States

I'd recommend approaching a local statistician to help work out a statistical design and help with interpretation of results. I unfortunately don't have time to provide design consultation on the support site, but have to restrict myself to software related questions.

0
Entering edit mode

sorry about the earlier post, somehow I pushed the wrong combination of keys and the post went online before I finished writing it. I understad, your time-limitation. Thanks anyway :)