View Single Post
Old 05-04-2017, 06:45 AM   #1
Neytiri
Junior Member
 
Location: London

Join Date: Nov 2012
Posts: 3
Default time course multi-factor design with DESeq2

Hi,

I'm looking for some help on the analysis of a multi-factor design with DESeq2.

I have a 4-factor experiment:
Groups: A (8 Donors), B (4 Donors), C (4 Donors), D (4 Donors)
Treatment: Control vs Infection
Time: 8h, 16h, 24h

The approach I took was combining all 3 factors (Group, Treatment, Time) into a single factor and using design=~ Donor + Group_Treatment_Time.

Then I looped through time points and through treatments to get differential expressed genes between two groups for each treatment and time-point:

res <- results(dds, c("Group_Treatment_Time", paste("A", treatment, timepoint, sep="-"), paste("B", treatment, timepoint, sep="-")))

However, I want to take advantage of the time course. In particular, I'm interested in time-specific differences:
i) between control and infection for each group
ii) between groups for control (or infection)

and I'm interested in finding genes with a difference in baseline expression (a main effect), ie lines moving in parallel:
iii) between control and infection for each group
iv) between groups for control (or infection)

My questions are:

1) Should I use a different design and LRT to find all genes with time-specific differences? How should that design be with 4 different factors?

2) How should I set up the contrasts for the cases i) to iv)?

Thank you so much for your help! I've read the vignette, many threads on the web, and talked to a couple of colleagues but I am still confused about how to run this analysis.

Regards,
Ana

Last edited by Neytiri; 05-04-2017 at 06:53 AM.
Neytiri is offline   Reply With Quote