Exploratory data anaylsis to detect batch effects

Question

[serpalma.v]

Dear members:

I am new at microarray data anaylsis. I am inspecting the log2 transformed data from an affimetrix array and this is the boxplot.

From what I have read, there seems to be no batch effect since the medians and means are very similar:

https://drive.google.com/file/d/0B82aQne0d_mYX1BGWHhHblgzNUU/view?usp=sharing

How would you interpret this plot and how else can I detect a batch effect?

Thanks for the help!

Answer 1

[h.mon]

Without labels, it is impossible to say anything meaningful about your data. I don't think boxplots are a good way to explore batch effects, a better way to explore the data are PCA (principal component analysis) or MDS (multidimensional scaling) plots. limma has a function for MDS plots.

Answer 2

[trausch]

There is a free HarvardX course on data analysis in the life sciences:

https://github.com/genomicsclass/labs

It has lots of exercises in R with real microarray & NGS data and has a full chapter on detecting batch effects and exploratory data analysis (chapter 2 & 10):

http://genomicsclass.github.io/book/