Principal components analysis of control kidney and liver data. Principal components were computed for normalized intensities from (A) kidney and (B) liver samples run on RAE230 2.0 arrays. The first two principal components are shown, which together explain about 80% and 53% of the total variation for kidney and liver, respectively. Each point in the plot represents an individual sample. (A) For kidney, organ section (indicated by color) appears to be the prominent explanatory factor. Lab is indicated by marker shape and gender is indicated by marker size. (B) Lab, diet, and fasting (indicated by color, shape, and size, respectively) are confounded in the liver dataset, and their respective levels are strongly associated with three distinct clusters. The sources and numbers of the 3 rat chows used in this dataset were Purina 5002, National Toxicology Program (NTP) 2000, and Provimi Kliba (PK) 3893.