Multi-factorial Time-based design in DEseq2: defining full and reduced model
1
0
Entering edit mode
Alex • 0
@da74b9f0
Last seen 9 days ago
United States

Hi,

My design looks like this ==I have 3 factors:

I'm interested to look at the effects of treatment(infection) on the gene expression in each organ over time. I'm having troubles understanding how to define the reduced model to run DESeq.

here is my code:

dds <- nbinomLRT(dds, full = ~Treatment + Organ + Time + Treatment:Time + Organ:Time, reduced = ~Treatment + Organ + Treatment:Time)

#rlog transformation
rld <- rlog(dds, blind=FALSE)
rld_mat <- assay(rld)

My questions:

  1. Are my full and reduced models correct for the intended purpose?
  2. Is my code for rlog transformation correct with "blind=FALSE" or should I use "blind=TRUE"?

Thanks!

DESeq2 • 217 views
ADD COMMENT
0
Entering edit mode
@mikelove
Last seen 2 days ago
United States

For questions about how to formulate particular contrasts in R, I recommend working with a local statistician or someone familiar with linear models in R. I have limited time on the support site, and have to reserve my time for software related questions.

It is fine to use blind=FALSE.

ADD COMMENT
0
Entering edit mode

Thanks.

ADD REPLY
0
Entering edit mode

Hi I was able to get an insight on the design. What is the difference in hypothesis testing between these two designs?

1. nbinomLRT(dds2, full = ~Organ + Time + Organ:Time, nbinomLRT(dds, full = ~Organ + Time + Organ:Time, reduced = ~Organ + Time)
2. nbinomLRT(dds2, full = ~Organ + Time + Organ:Time, reduced = ~1)

when I run both, there is a difference in the number of genes that are differentially expressed.

summary(res_LRT) ====1st design! 

out of 1359 with nonzero total read count
adjusted p-value < 0.1
LFC > 0 (up)       : 11, 0.81%
LFC < 0 (down)     : 8, 0.59%
outliers [1]       : 0, 0%
low counts [2]     : 869, 64%
(mean count < 2)
[1] see 'cooksCutoff' argument of ?results
[2] see 'independentFiltering' argument of ?results

and

summary(res_LRT2) ====2nd design!

out of 1359 with nonzero total read count
adjusted p-value < 0.1
LFC > 0 (up)       : 56, 4.1%
LFC < 0 (down)     : 67, 4.9%
outliers [1]       : 0, 0%
low counts [2]     : 527, 39%
(mean count < 1)
[1] see 'cooksCutoff' argument of ?results
[2] see 'independentFiltering' argument of ?results
ADD REPLY
0
Entering edit mode

For questions about statistical analysis and interpretation, I recommend working with a local statistician or someone familiar with linear models in R.

ADD REPLY

Login before adding your answer.

Traffic: 354 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

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

Powered by the version 2.3.6