DESeq2 timecourse experiments
1
0
Entering edit mode
@sconosciuto-24102
Last seen 7 months ago

Hi everybody,

I have done a DESeq2 analysis using the package's vignette, anyway I have a some points I cant understand: I have 2 different conditions (wt vs mut) and 7 different time points (T0 to T6)

I obtained the genes with greatest differences over the time points, ordered by adj. p-value:

ddsTC = DESeqDataSetFromTximport(txi, design = ~ condition + timepoints + condition:timepoints, colData = dfSamples)
ddsTC = DESeq(ddsTC, test="LRT", reduced = ~ condition + timepoints)
resTC = results(ddsTC)


Then, do I obtain the genes with greatest differences between T1 with respect to T0 using commands? And what exactly it means? Are these the genes which have a similar expression at T0 and an increased difference at T1 or are these the genes with an high difference both at T0 and T1?

resultsNames(ddsTC)

[1] "Intercept"               "condition_mut_vs_wt"
[3] "timepoints_t1_vs_t0"     "timepoints_t2_vs_t0"
[5] "timepoints_t3_vs_t0"     "timepoints_t4_vs_t0"
[7] "timepoints_t5_vs_t0"     "timepoints_t6_vs_t0"
[9] "conditionmut.timepointst1" "conditionmut.timepointst2"
[11] "conditionmut.timepointst3" "conditionmut.timepointst4"
[13] "conditionmut.timepointst5" "conditionmut.timepointst6"

resTC_t1vst0 = results(ddsTC, name="timepoints_t1_vs_t0")


In case I would compare samples at T0 as done in a classical analysis (3wt vs 3mut) is there a faster way instead of load with only one time point with tximport ?

Thanks

deseq2 timecourse • 148 views
1
Entering edit mode
@mikelove
Last seen 3 hours ago
United States

"And what exactly it means? Are these the genes which have a similar expression at T0 and an increased difference at T1 or are these the genes with an high difference both at T0 and T1?"

There is some explanation in the workflow, but also in the vignette section on interactions. If after reading over the material, the interaction designs are still not clear, I would recommend approaching a local statistician who you could work with to interpret these terms.

"In case I would compare samples at T0 as done in a classical analysis (3wt vs 3mut) is there a faster way instead of load with only one time point with tximport ?"

It's preferable to fit one model above and extract comparisons as needed.