Analyzing FACS Data I have three FACS cell sorting datasets and I'm trying to find the best statistical test. The problem is that the data is in percentages, and I've spent all month searching for a test with no luck.
Sample         GreenCells    RedCells   BlueCells  TotalCellsSorted
 1                  4500      4600          1200      19000
 2                  6400      11500         1800      34000
 3                  6400      6500          1700      25000

I want to prove statistically that the red cells vary significantly, and the blue and green cells remain relatively consistent. Total cells vary due to the fact they may be lost during the separation process. 
Any help would be appreciated.
 A: With the absolute numbers given, we can see the experiment as a number of binomial decisions. Instead of percentages we examine the probabilities behind each observations in a binomial model. You now need do define in more detail, what

that the green cells vary significantly, and the red and blue cells remain relatively consistent

exactly means. What would persuade people to believe that. credible values for each of the probabilies could be visualized like this:
par(mfrow=c(3,1), mar=c(3,2,1,1))
curve(dbeta(x, 4500, 19000-4500), main = "Green cells", xlim=c(0,1), ylim=c(0,180))
curve(dbeta(x, 6400, 34000-6400), add= TRUE)
curve(dbeta(x, 6400, 25000-6400), add= TRUE)

curve(dbeta(x, 4600, 19000-4600), main = "Red cells", xlim=c(0,1), ylim=c(0,250))
curve(dbeta(x, 11500, 34000-11500), add= TRUE)
curve(dbeta(x, 65400, 25000-6500), add= TRUE)

curve(dbeta(x, 1200, 19000-1200), main = "Blue cells", xlim=c(0,1), ylim=c(0,180))
curve(dbeta(x, 1800, 34000-1800), add= TRUE)
curve(dbeta(x, 1700, 25000-1700), add= TRUE)


I feel this is proof enough, for how you worded the problem. If you want a formal test, we need a more precise wording, what to test.
