I have created a de novo RNA-seq assembly using Trinity with samples from 4 treatment groups and then ran a DEG analysis on each pairwise combination of the 4 groups. Then, I looked at how DEGs that overlapped between two of the comparisons were correlated. I used a couple of different ways to do this but one of the ways was plotting a scatterplot of the log 2FC from one comparison vs the log2FC from the other comparison. 
In this scatterplot I have multiple diagonal lines of points running through the plot. When I investigate in more detail it looks like there are some genes with very similar log2 FCs (like 8.1.., 8.12..., 8.23.., etc). My two potential suspicions are that maybe it has something to do with low gene counts or something of the sort since I have quite a lot of Trinity "genes" but still I filtered for low CPM in the DEG analysis. Or maybe it has something to do with the fact that the two comparisons I am comparing are Group 1 vs Group 2 and Group 3 vs Group 2. Any suggestions you have would be greatly appreciated!
$\begingroup$
$\endgroup$
1
-
1$\begingroup$ (Biological ignoramus here.) I am guessing that each coordinate is the log of a ratio, so of the form log (u / w) or log( v / w). If you can isolate the most obvious linear cluster and show us not only the coordinates as plotted but also the original data, you might get closer to an answer. $\endgroup$– Nick CoxCommented Mar 4 at 17:13
Add a comment
|