Question: Deseq2 design for one condition over 8 different time point
0
gravatar for thindmarsmission
14 months ago by
thindmarsmission0 wrote:

HI! I have a question, about design matrix in Deseq2, when I don't have any control or conditions. But same sample at different time points for differential expression analysis . I want to do differential expression between different time points. How I should define my Design ?

design(dds) <- ~Time
dds <- DESeq(dds, test="LRT")
res <- results(dds)

 

Also, in another case if I have 3 biological replicates for each time point [no treatment/condition] and want to analyse across time points . It should be like:

design(dds) <- ~replicates + time
dds <- DESeq(dds, reduced=~replicates, test="LRT")
res <- results(dds)

Is it correct way?

I was trying to take a look o  another question, but in that case they have conditions.

DESeq2 likelihood ratio test (LRT) design - 2 genotypes, 4 time points

ADD COMMENTlink modified 14 months ago by Michael Love26k • written 14 months ago by thindmarsmission0
Answer: Deseq2 design for one condition over 8 different time point
2
gravatar for Michael Love
14 months ago by
Michael Love26k
United States
Michael Love26k wrote:

If you have replicates that track across time you would use ~rep + time for the full model and ~rep for the reduced model. If you don't have replicates that track across time, you would use ~time for the full model and ~1 for the reduced model.

ADD COMMENTlink written 14 months ago by Michael Love26k

Thanks for your reply Michael,

yes, in one case i have replicates across all time points, I used ~rep+time for full design and ~rep for reduced. I got following results

> resultsNames(dds)
[1] "Intercept" "rep"       "time"    

Should not I get, like "t1_vs_t2" "t2_vs_t3" etc.. actually, I wanna understand how to interpret it? any suggestions for articles and links.  thanks again.

ADD REPLYlink written 14 months ago by thindmarsmission0
1

There is a message that DESeq2 prints out, that you may have missed, that tells you about options for modeling numeric covariates.

If you want to treat time as a categorical you need to run factor() on the variable first.

ADD REPLYlink written 14 months ago by Michael Love26k

thanks, i got following combinations:(and sorry to bother you again)

> resultsNames(dds)
 [1] "Intercept"     "rep_a2_vs_a1"  "rep_a3_vs_a1"  "time_t2_vs_t1" "time_t3_vs_t1" "time_t4_vs_t1" "time_t5_vs_t1"  "time_t6_vs_t1" "time_t7_vs_t1" "time_t8_vs_t1"

What is the meaning of "reduced=~rep" and if I want combination of all the time points, what design should i use,(here I can find combinations only with t1 and rest of time points).

Thanks a lot.

ADD REPLYlink written 14 months ago by thindmarsmission0
1
Take a look at the workflow (linked from the beginning of the vignette). It has a time series example.
ADD REPLYlink modified 14 months ago • written 14 months ago by Michael Love26k

Dear Love,

I would like to put you a question related to the design of an RNA-seq experiment. If I have 65 samples and of them, 7 samples are from one condition and the rest from the other condition, could I perform a DE analysis with the total amount of samples in Deseq2 or should I select a subset of 7 samples from those 58 samples from one of the two conditions?

Thank you in advanced.

ADD REPLYlink written 14 months ago by Teresa0

This is a FAQ in the DESeq2 vignette.

ADD REPLYlink written 14 months ago by Michael Love26k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 383 users visited in the last hour