Question

Microarray Data - Pca On Anova

1

Entering edit mode

11.9 years ago

James ▴ 80

Hi All,

Does anyone know how to do PCA analysis on microarray data on which ANOVA has already been carried out, using R or Matlab?

Thanks

James

pca r • 2.7k views

ADD COMMENT • link updated 11.9 years ago by Sean Davis 26k • written 11.9 years ago by James ▴ 80

1

Entering edit mode

Not sure what you want to do. The input to PCA is typically normalized microarray data. How does the ANOVA come into it?

ADD REPLY • link 11.9 years ago by Sean Davis 26k

0

Entering edit mode

I'm trying to do something like that too. What happened finally ? Did you run PCA on your anova results ?

ADD REPLY • link 7.0 years ago by arronar ▴ 280

score 0 · Answer 1 · 2012-06-15

More details:

I actually have a data set as follows:

rows: a list of 20, 000 genes

columns: expt1 expt2 expt3 expt4 expt5

My aim was to apply one-way anova on this data so as to obtain lists of significant genes in the 5 experiments. I will then apply PCA on the selected genes. My problem is how to achieve these steps using R. I will be happy to get some advice.

Thanks

James

score 0 · Answer 2 · 2012-06-18

See the limma package from Bioconductor. It has an excellent user guide.

As for PCA after differential expression, that is fine to do (see the prcomp function, for example), but one needs to realize that doing so will often result in the samples "clustering" together into expected groups. This is NOT evidence of sample similarity since the PCA used a set of genes for input that were chosen to maximize differences between groups.