Question

Time points and donors - errors with contrasts

0

Entering edit mode

sasha • 0

@sasha-11847

Last seen 7.4 years ago

Dear community,

I've been trying to test if there are differentially expressed genes between different time-points of samples taken from the same donors.

Experiment:

SampleID Donor TimePoint

A1 1 T1

A2 1 T2

A3 1 T3

A4 2 T1

A5 2 T2

A6 2 T3

And so it goes until sampleID A12, so in total I have 4 different Donors and for each I have 3 timepoints.

I used a design that takes into account Donor (given that the samples from the same donor are not independent) and TimePoint: ~ Donor + TimePoint. I then used test="LRT" when calling DESeq function with a reduced design: ~Donor to find genes which are different at any time-point. After that I tried to test individual effects by specifying a contrast in the results function:

Res <- results(DDS, contrast=c("TimePoint","T1","T2"))

But it give me an error:

Error in cleanContrast(object, contrast, expanded = isExpanded, listValues = listValues,  : 
  as T2 is the reference level, was expecting TimePoint_T1_vs_T2 to be present in 'resultsNames(object)'

I also tried:

Res <- results(DDS, contrast=c("TimePoint","T1","T2"), test="Wald")

And it gave me the same error. What am I doing wrong? Should I do this differently?

From just running results without contrast: Res <- results(DDS) I get no errors and the results look as follows:

log2 fold change (MLE): TimePoint2 
LRT p-value: '~ Mother_Child + TimePoint' vs '~ Mother_Child' 
DataFrame with 460 rows and 6 columns
          baseMean log2FoldChange     lfcSE      stat      pvalue      padj
         <numeric>      <numeric> <numeric> <numeric>   <numeric> <numeric>
Otu402    2.279747      0.3111686 0.8422597 0.1381527   0.9332554      0.043
Otu131   22.551497      2.0865825 0.8764297 3.3208392   0.1900592      0.032

Why does it say "TimePoint2"? I thought it was the result of comparing T1 vs T2, because is using as reference TimePoint T1, but when I try relevel to set T2 as reference, the results are exactly the same :(

Please, any help would be greatly appreciated :)

Cheers

Sasha

deseq2 contrast • 1.9k views

ADD COMMENT • link updated 7.0 years ago by Michael Love 43k • written 7.4 years ago by sasha • 0

0

Entering edit mode

Dear Michael,

Thank you for your reply. I checked the character string as you suggested:

> levels(DDS$TimePoint)
[1] "T2" "T1" "T3"

> resultsNames(DDS)
[1] "Intercept"     "Mother_Child1" "Mother_Child2" "Mother_Child3" "TimePoint1"      "TimePoint2"

Just to let you know, Mother_Child = Donor in my case (forgot to state that).

Still don't understand what am I doing wrong. Maybe there's something else I could try changing?

Thanks in advance

Sasha

ADD REPLY • link updated 7.4 years ago by Michael Love 43k • written 7.4 years ago by sasha • 0

0

Entering edit mode

Can you show all the code from scratch? It should just give the levels of the variable in resultsNames(), unless you change them after running DESeq()

ADD REPLY • link 7.4 years ago Michael Love 43k

score 0 · Answer 1 · 2017-07-12

0

Entering edit mode

Michael Love 43k

@mikelove

Last seen 1 day ago

United States

You have to be careful that you specify exactly the right character strings you used for the factor levels.

You could check

resultsNames(dds)

or also

levels(dds$TimePoint)

To see what those are

ADD COMMENT • link 7.4 years ago Michael Love 43k

score 0 · Answer 2 · 2017-07-20

Hi Michael,

I realised just recently that the problem is related to the "SAN controllers" from the unix system (where I use R studio from) in my institute. They are actually having big problems at the moment. When I repeated everything in my R studio (in my personal computer) everything worked perfectly.

Thanks in advance and sorry for the inconvenience,

Alexandra