Principal component analysis of select samples. Data is filtered by the Coefficient of Variation. Samples are plotted by expression of the first two Principal Components. Key: mild disease (blue); severe disease (red); training set (open circles), validation set (closed squares). (A) Severity of disease is assessed by FVC% (see text): identified 31 cases with mild disease and 10 cases with severe disease. (B) Severity of disease is assessed by DLCO%: 14 cases of mild disease versus 18 cases of severe disease.