Dear Community,
I am working on a three-way cross species project comparing differential expression between three different species & two different tissues. Therefore, i have performed a hierarchical cluster analysis using "hclust" in R of mapped reads normalized log2(cpm+1). However, the result is contradicting:
a) DEG results (DESEq2; more difference between clustered species than unclustered (higher divergence)).
Since one of my species has ~13% lower mapping efficiency (TOPHAT2) than the others. And exactly this one is clustering with highest divergence to the others. I am questioning, if the mapping efficiency could have an impact and can be normalized?
Thanks for your help!
Hi Björn,
You can try to assess the clustering of your data before and after upper-quartile normalization to check if your weird clustering is due to a difference in reads mapped in your different samples. I use the EDAseq package (http://bioconductor.org/packages/release/bioc/vignettes/EDASeq/inst/doc/EDASeq.html#introduction). You can feed in directly the count file (usually called cts.txt) derived from DESeq/DESeq2.
Here the script I used:
```{r}
library(EDASeq)
# Library for plotting
library(RColorBrewer)
colors
Thank you very much for your answer. I tried it by using riog normalized data now. It worked. rlog is a function within DEseq2 used to transforms the count data to the log2 scale in a way which minimizes differences between samples for rows with small counts, and which normalizes with respect to library size.
However, I will give EDAseq a try in the future.
How about the method (Single, Average and complete) that used in your clustering , have you used different one of them to conduct the result.
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping