Tweet
Hi all,
I've got a relatively complex RNA-Seq experiment and I was hoping to get some input on analysis options. I've dug around quite extensively in these forums and others, but haven't been able to find an experimental design that's quite the same.
Essentially, I have a set up which has 2 groups based on the outcome of an infection (e.g. no-complications vs. complications). For each group, I have 4 time points (e.g. pre-infection, early infection, late infection, very late infection). So, the data are essentially "paired" - but rather than a single pair (e.g. pre-infection vs. post-infection), they've got multiple time points. For each group-timepoint, I have 10 biological replicates for RNA-Seq on tissue samples.
Example:
-No-complications
Patient 1: pre, early, late, very late
Patient 2: pre, early, late, very late
…
Patient 10: pre, early, late, very late
-Complications
Patient 1: pre, early, late, very late
Patient 2: pre, early, late, very late
…
Patient 10: pre, early, late, very late
Though there are several, the primary objective is to observe gene expression differences in the no-complications vs. complications group. This would be relatively straightforward for a traditional paired experiment using a GLM-type of analysis. In fact, I plan to start by using DESeq with GLM analysis JUST comparing the first 2 time points (pre vs. early). However, it would be great to take advantage of the additional time course data in a unified analysis.
Other than clustering/heatmap types of data exploration, can anyone offer additional advice on how to manage this type of design from an RNA-Seq perspective? There are a few microarray-type tools available, but I don't think they're well tailored to these type of data.
Any advice would be much appreciated. Thank you in advance.
I've got a relatively complex RNA-Seq experiment and I was hoping to get some input on analysis options. I've dug around quite extensively in these forums and others, but haven't been able to find an experimental design that's quite the same.
Essentially, I have a set up which has 2 groups based on the outcome of an infection (e.g. no-complications vs. complications). For each group, I have 4 time points (e.g. pre-infection, early infection, late infection, very late infection). So, the data are essentially "paired" - but rather than a single pair (e.g. pre-infection vs. post-infection), they've got multiple time points. For each group-timepoint, I have 10 biological replicates for RNA-Seq on tissue samples.
Example:
-No-complications
Patient 1: pre, early, late, very late
Patient 2: pre, early, late, very late
…
Patient 10: pre, early, late, very late
-Complications
Patient 1: pre, early, late, very late
Patient 2: pre, early, late, very late
…
Patient 10: pre, early, late, very late
Though there are several, the primary objective is to observe gene expression differences in the no-complications vs. complications group. This would be relatively straightforward for a traditional paired experiment using a GLM-type of analysis. In fact, I plan to start by using DESeq with GLM analysis JUST comparing the first 2 time points (pre vs. early). However, it would be great to take advantage of the additional time course data in a unified analysis.
Other than clustering/heatmap types of data exploration, can anyone offer additional advice on how to manage this type of design from an RNA-Seq perspective? There are a few microarray-type tools available, but I don't think they're well tailored to these type of data.
Any advice would be much appreciated. Thank you in advance.