Hello
I have a dataset composed by a time series of 4-time points, two treatments, and a time zero control.
My interest is to identify the differentially expressed genes in each treatment with respect to the different time points. To achieve this, I compared first the treatments against the control, however, I am also interested in identifying those DEG between each time point and each treatment. Is there a way to include both comparisons in the same design model? This is my dataset description and what I have done so far:
``
Inv Tratamient Day Name
1 1A control 0T EA2SS13.genes.results
2 1A control 0T EA2SS22.genes.results
3 1A control 0T EA2SS31.genes.results
4 2B p 1T EA2SS18.genes.results
5 2B p 1T EA2SS27.genes.results
6 2B p 1T EA2SS36.genes.results
7 3C p 3T EA2SS19.genes.results
8 3C p 3T EA2SS28.genes.results
9 3C p 3T EA2SS37.genes.results
10 4D p 6T EA2SS20.genes.results
11 4D p 6T EA2SS29.genes.results
12 4D p 6T EA2SS38.genes.results
13 5E p 9T EA2SS21.genes.results
14 5E p 9T EA2SS30.genes.results
15 5E p 9T EA2SS39.genes.results
16 6F o 1T EA2SS14.genes.results
17 6F o 1T EA2SS23.genes.results
18 6F o 1T EA2SS32.genes.results
19 7G o 3T EA2SS15.genes.results
20 7G o 3T EA2SS24.genes.results
21 7G o 3T EA2SS33.genes.results
22 8H o 6T EA2SS16.genes.results
23 8H o 6T EA2SS25.genes.results
24 8H o 6T EA2SS34.genes.results
25 9I o 9T EA2SS17.genes.results
26 9I o 9T EA2SS26.genes.results
27 9I o 9T EA2SS35.genes.results
``
What I have been doing is to follow the interaction section of the DESeq2 vignette:
```dds <- DESeqDataSetFromTximport(txi, samples, ~1)
colData(dds)
dds$group <- factor(paste0(dds$Tratamient, dds$Day))
design(dds) <- ~ group
dds <- DESeq(dds)
resultsNames(dds)```
As a result, I got the comparisons of each day and treatment against the control:
```resultsNames(dds)
[1] "Intercept" "group_o1T_vs_control0T" "group_o3T_vs_control0T" "group_o6T_vs_control0T"
[5] "group_o9T_vs_control0T" "group_p1T_vs_control0T" "group_p3T_vs_control0T" "group_pT_vs_control0T"
[9] "group_p9T_vs_control0T"```
However, I want to know the influence of the treatment over time. Do you have any suggestion?
Also, I followed the suggestion of Deseq2 - Time-series data with 2 treatments and 1 control. Here it was suggested to use the days separately, which I did too:
```colData(dds)
DataFrame with 9 rows and 3 columns
Day Tratamient Name
<factor> <factor> <factor>
EI1TA2SS13 T0c control EA2SS13.genes.results
EI1TA2SS22 T0c control EA2SS22.genes.results
EI1TA2SS31 T0c control EA2SS31.genes.results
EI1TA2SS18 T1p p EA2SS18.genes.results
EI1TA2SS27 T1p p EA2SS27.genes.results
EI1TA2SS36 T1p p EA2SS36.genes.results
EI1TA2SS14 T1o o EA2SS14.genes.results
EI1TA2SS23 T1o o EA2SS23.genes.results
EI1TA2SS32 T1o o EA2SS32.genes.results```
dds <- DESeqDataSetFromTximport(txi, samples, ~1)
colData(dds)
design(dds) <- ~ Tratamient
dds <- DESeq(dds)
resB <- results(dds, contrast=c("Tratamient", "o", "control"))
resBA <- results(dds, contrast=c("Tratamient", "o", "p"))
However, I have doubts on how to interpret these results. Is it valid to merge both results (resB and resBA) to see all the differentially expressed genes in one particular condition (o in this case) against treatments and controls? Or should I do the analysis always separately?
