RNAseq time series data w/ controls for each time point

Mocca · 08-07-2013, 05:32 AM

Hi everyone,

I am currently working on a RNAseq data set consisting of 4 time points where each time point has 6 challenged and 6 control individuals (not paired). I have used both DEseq and edgeR for pairwise comparisons such as: time 1 - challenged vs control which worked nicely.

I am not sure that the reported DE genes at one time point is comparable to the DE genes at another time point. Thus I am looking for a package that can handle time series data. I have found several but none of them handles controls on each time point but prefers a base line expression level or a time 0 (which I don't have).

I can manage with my pairwise analyses, but if any of you have good suggestions to what I might try I would appreciate it greatly!

Thanks in advance

dpryan · 08-07-2013, 05:53 AM

You could use DESeq or edgeR still and just not do the pair-wise comparisons. Using DESeq syntax:

Code:

fit1 <- fitNbinomGLMs(cds, count~treatment*time)
fit0 <- fitNbinomGLMs(cds, count~time + treatment:time)
pvals <- nbinomGLMTest(fit1, fit0)

...assuming you're interested in genes that are differentially expressed by treatment when accounting for time point and a possible time-point:treatment interaction.

Mocca · 08-07-2013, 11:12 PM

Thanks for the reply.

I didn't consider doing it this way. Got a bit caught up in my control samples i guess. Will give it a try!

Cheers

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
RNAseq time course data	rmberka	RNA Sequencing	7	12-16-2013 07:03 AM
Comparative Time Series Analysis of RNA-Seq data?	ashuchawla	Bioinformatics	8	07-25-2013 12:12 PM
can DESeq handle time series data now?	Jeremy	RNA Sequencing	0	12-17-2012 09:15 PM
How to analyse time-series data with edgeR	ca85	RNA Sequencing	0	10-06-2011 05:04 AM
time series analysis of count data?	greigite	Bioinformatics	0	06-14-2011 05:11 PM

08-07-2013, 05:32 AM	#1
Mocca Junior Member Location: Oslo Join Date: Mar 2012 Posts: 9	RNAseq time series data w/ controls for each time point Hi everyone, I am currently working on a RNAseq data set consisting of 4 time points where each time point has 6 challenged and 6 control individuals (not paired). I have used both DEseq and edgeR for pairwise comparisons such as: time 1 - challenged vs control which worked nicely. I am not sure that the reported DE genes at one time point is comparable to the DE genes at another time point. Thus I am looking for a package that can handle time series data. I have found several but none of them handles controls on each time point but prefers a base line expression level or a time 0 (which I don't have). I can manage with my pairwise analyses, but if any of you have good suggestions to what I might try I would appreciate it greatly! Thanks in advance

08-07-2013, 05:53 AM	#2
dpryan Devon Ryan Location: Freiburg, Germany Join Date: Jul 2011 Posts: 3,480	You could use DESeq or edgeR still and just not do the pair-wise comparisons. Using DESeq syntax: Code: fit1 <- fitNbinomGLMs(cds, count~treatmenttime) fit0 <- fitNbinomGLMs(cds, count~time + treatment:time) pvals <- nbinomGLMTest(fit1, fit0) ...assuming you're interested in genes that are differentially expressed by treatment when accounting for time point and* a possible time-point:treatment interaction.

08-07-2013, 11:12 PM	#3
Mocca Junior Member Location: Oslo Join Date: Mar 2012 Posts: 9	Thanks for the reply. I didn't consider doing it this way. Got a bit caught up in my control samples i guess. Will give it a try! Cheers