Hi,
I was wondering if there are some tools able to model DEG / pathway enrichment in bulk RNA-seq when using fast-dividing cancer cell lines.
In the current analysis I am working, we are testing a couple of cancer therapy compounds vs controls at 2 timepoints. I've used the standard pipeline of DESeq2 + ORA/GSEA. As you can imagine, we are interested in (and already detecting) pathways related to upregulation ofstress/ damage response, and cell death, and downregulation of cell cycle, cell division, etc.
However, the control cell line, at the 2nd timepoint, it has divided so much that the cells are mostly stopping dividing (downreg of anything cell cycle, cell division, proliferation,...) , despite not having reached 100% confluence yet. This makes any comparison between treatments and controls at the 2nd timepoint unreliable, because they show upregulation of cell cycle, just because the controls stopped.
Still, we all know that the transcriptomic profile of cells in a plate change with time, just by sitting there. There are plenty of genes showing identical patterns in both controls and treatments just by the passing of time. So what I have is a combination of effects:
- cells just sitting on a plate by time (both in ctrls and treatment) <---- I don't care
- therapy by time (only treatment) <---- what we are interested in
- controls replicating so fast that they stopped dividing (only controls) <---- This messes up with the analysis
I am already analyzing the simple pairwise comparison of treatmentA/B-time2 vs treatmentA/B-time1, as well as a treatmentB-time2 vs treatmentA-time2 and an interaction model with design=treatment*time using treatmentA as the baseline instead of controls. However, there are almost no differences between treatments. I'm interested in seeing if I can dig a bit deeper in the time2 vs time1 comparisons.
Do you know of any (statistically sound) method to decouple these effects from the samples to try to find the actual effects of therapy?
On top of that, all the pathways affected (death, damage/stress response, replication, repair, cell cycle/division/proliferation) are very closely related and share a ton of genes. Are there methods capable of deconvoluting the weight/effect of each pathway?
Thanks
