PCA and Hierarchical Clustering of One Cohort using 100-Probeset Signature Generated from Second Cohort. Plots of first versus second component of PCA shows separation of non-molecular apocrine (blue) samples from molecular apocrine (green) samples. Dendrograms denote phenotype as 0 preceding sample number for "non-molecular apocrine" and 1 preceding sample number for "molecular apocrine". Data presented in (A) uses gene signature from Farmer et al to perform PCA and HC on Doane cohort and data presented in (B) uses gene signature from Doane et al to perform PCA and HC on Farmer cohort. All data was log transformed, quantile normalized, processed by XPN, and used AffyProbeMiner CDF.
Sanga et al. BMC Medical Genomics 2009 2:59 doi:10.1186/1755-8794-2-59