Entering edit mode
4.0 years ago
Lorenzo
▴
10
Hi to all,
I obtain this plot after vst method of DESeq2. 
Should I proceed with batch effect correction or my data are good without also without the correction?
Sorry for my question but I'm not sure how to interpret this plot.
Thank you in advance for any suggestion!
What you see here in the PCA plot is how similar (or different) your samples are.
You should question yourself whether your samples are clustered by batch or by biological semblance. If the clustering you see has sense biologically, then you can proceed with the analysis, otherwise you should do a batch correction.
I was hoping that someone could answer your statement by looking at the plot ,if possible, I mean how can I understand the meaning of the cluster? The batch that I could consider could be the different days of library preparation (e.g Date1 and Date2 in the plot).
Color the points by Date only, so it will be easier to interpret if certain dates make up certain clusters. It can be good to look at further dimensions if they explain a relevant % of the variability.
You can do this visual approach for any variables which you suspect influence the variability in your data. For a more "automated" approach, you could calculate the correlation between PCA variables and your experimental variables.
Thank you for your answer. This is the plot by date only.
What do you mean with "look at further dimensions"?
Look at other PCs. In this plot you are representing PC1 and PC2. You could instead represent PCs 3,4, etc. It looks that the day the different samples were processed is not affecting your analysis. However, I would do the same plot with cell type instead to be entirely sure.
Ok thank you very much for your suggestions!