Comparison of methods to identify aberrant expression patterns in individual patients: augmenting our toolkit for precision medicine

• DOI: 10.1186/gm509
• Publication Date: 2013-11-29T02:01:35Z
• Authors: Daniel Bottomly; Peter A Ryabinin; Jeffrey W Tyner; Bill H Chang; Marc M Loriaux; Brian J Druker; Shannon K McWeeney; Beth Wilmot
• Keywords: Human genetics; Expression (computer science)

Patient-specific aberrant expression patterns in conjunction with functional screening assays can guide elucidation of the cancer genome architecture and identification therapeutic targets. Since most statistical methods for analysis are focused on differences between experimental groups, performance approaches patient-specific analyses currently less well characterized. A comparison genes that dysregulated relative to a single sample given set samples, our knowledge, has not been performed. We systematically evaluated several including variations nearest neighbor based outlying degree method, as Zscore robust variant their suitability detect events. The were assessed using both simulations data from cohort pediatric acute B lymphoblastic leukemia patients. first power false discovery rates found even under optimal conditions, high effect sizes (>4 unit differences) necessary have acceptable any method (>0.9) though (>0.1) pervasive across simulation conditions. Next we introduced technical factor into was reduced all weights could provide gains depending number samples affected by factor. In use case highlights integration patient (the gene dysregulation events associated targets siRNA screen), demonstrated successfully identify one sample. However, only samples. Our results show may be useful alternative or Rscore personalized medicine context especially small medium sized (between 10 50 samples) datasets moderate sample-to-sample variability. From these guidelines detection precision context.