batch effects

[fabpitt]

Hello, I'm a student.
I have an expression matrix of 1208 samples (1095 tumor and 113 normal) downloaded from TCGA (RNA-Seq) and I have to adjust for batch effects. http://bioinformatics.mdanderson.org/tcgambatch/ suggests that there are 3 batches: type(tumor/normal), plateId and TSS but in a different forum, a boy said me that he has never noticed these batch effects.
So I'm a bit confused...
Is there anyone who can explain me how to find batch effects? I know that I can look for them with PCA but I have no idea how to interpret the graph. Can I consider tumor and normal as batch effects?
Thanks in advance

[Dario1984]

No; tumour and normal samples are not unwanted technical variation. They are phenotypes of biological interest. A lot of analyses have been done with TCGA data and published, so I think you could ignore batch effects for your analysis as everyone else has done. If you don't understand a PCA plot, then you shouldn't try to estimate batch effects, which is much more complicated to do correctly.