I am currently trying to run WGCNA for my microarray dataset. I have run QC checks and removed outliers, then I have normalised using rma() and finally when I run the code to make a sample tree I had 1 outlier. I then removed that outlier manually and run the code from the start with the samples without the outlier sample. However when I go to make the sample tree again there is always another outlier. What is happening?

I am up to the hclust() step in the WGCNA data input tutorial.

What you are interpreting as an outlier may, in fact, not be an outlier. You have not shown any graphic or provided any other information such that we can independently check. In hierarchical clustering, there will virtually always be at least one sample that separates from a high-level branching structure into its own node - this is normal, and is not necessarily indicative of an outlier.

Please also check a PCA bi-plot of your samples, and note the percent explained variation on the axes for PC1 and PC2;

If you want, you can provide the GSE number of the study that you're analysing so that I can take a quick look at it.

Kevin