Hi there. If you're are testing the expression level of a certain protein and you would like to do statistics, you should do several replicates to increase the N. 

For western blot band quantifications, I use imagej and I quantify both the bands of interest and a household protein, beta-actin for example, to normalise for the relative protein expression. If you do this for several replicates you can have a mean and SD and then do statistics, the simplest one is students t test.