RNA-seq analysis without grouping data

Question

[e.antoun]

Hi,

I have some RNA-seq analysis that I want to do and determine differential gene expression with respect to several different phenotypic characteristics. For some of these, we can group the data (i.e. control, treatment), but we also want to determine differential gene expression with respect to several continuous variable such as fat mass and age.

Is there a way to determine differential gene expression in R without having to divide the samples into distinct groups, and to look at differential expression with respect to increasing age for example?

Thanks

[Nicolas Rosewick]

One idea could be to bin ages (or any other continuous variable) in discrete categories [0-5], [5-10], etc... and to use DESeq2 to perform a time-series analysis.

[e.antoun]

Thanks for the reply, that is one possibility but it is what we are hoping to avoid having to do, as it is technically still grouping the data. Obviously if there is no way around it, then we will have to go down that route,, but we are hoping to avoid that as much as possible.

Answer 1

[mark.ziemann]

The edgeR user guide has in depth description of GLM analyses (page 15) suitable for many types of analyses such as correcting for gender (factor) while looking for genes that correlate with age (continuous). I understand that DESeq provides more or less the same functionality too. NicoBxl's suggestion of stratifying individuals into groups is an excellent suggestion.

[e.antoun]

Thanks for the reply, I shall have a look at the GLM capabilities of edgeR and DESeq.