Interpretation of results names after LRT
Entering edit mode
A ▴ 40
Last seen 3 months ago
United Kingdom

Hi all, 

Just a quick question regarding a time point experiment I have done.  Having asked quite a few questions and read threads I think I have the right design and statistical test for looking for genes that change over time, however, I would just like confirmation if anybody can clarify, what the results output mean as it is concerning me slightly.

I have 10 organs and 4 time points. I want to look for changes across time in general. I don't want an interaction term as I am only interested in genes that change across time, and thats it. I will do gene patterns later on to look for similarities between organs etc... My question is, is the LRT test calculating changes over time for every single organ given the the results name output below. I have re-leveled for time point to be the first time point, but haven't re-leveled for a specific organ as I am only interested in the baseline being time 0 and genes that change after time 0. 

dds<-DESeqDataSetFromMatrix(countData = countdata1, colData = sampledata, design = ~ Time_point + Organ )

dds <- DESeq(dds, reduced = ~Organ, test = "LRT")

The results name is as follows:

[1] "Intercept"                      "Time_point_p15_vs_p3"          
 [3] "Time_point_p30_vs_p3"           "Time_point_p60_vs_p3"          
 [5] "Organ_Kidneys_vs_Brain"         "Organ_Large_Intestine_vs_Brain"
 [7] "Organ_Lungs_vs_Brain"           "Organ_Pancreas_vs_Brain"       
 [9] "Organ_Prostate_vs_Brain"        "Organ_Salivary_Glands_vs_Brain"
[11] "Organ_Small_Intestine_vs_Brain" "Organ_Stomach_vs_Brain"        
[13] "Organ_Uterus_vs_Brain"    

My question is: If the reduced of ~Organ looks specifically for changes after time 0, why am I getting results names for all organs vs the brain? I understand that its the brain, because by default the first factor for baseline is the organ in alphabetical order. Has the test looked for changes after 0 in all organs irrespective of any organ bias as would be suggested with the time point_p30_vs_p3 etc? and if so, what is the meaning of the organ comparisons?

Sorry if this is a stupid question but I really need this clarified as I'm confused by it and want to be absolutely sure of what the test is doing. 

Many thanks for your help!

DESeq2 LRT multiple time points • 857 views
Entering edit mode
Last seen 4 hours ago
United States

resultsNames() really returns all the coefficients, including the nuisance ones. So although you test for the time variable, the differences between organs are also present as coefficients. They aren't really "results" for you in this sense, but you could for example, look up the fitted differences between organs, because that is what happens when you have a design ~organ + time.

Entering edit mode

Thank you so much for clarifying!


Login before adding your answer.

Traffic: 775 users visited in the last hour
Help About
Access RSS

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

Powered by the version 2.3.6