Microarray Analysis of miRNA Expression

Total RNA was purified from cell lines using the miRNeasy kit (Qiagen, Valencia, CA, USA), and 100 ng was labeled and hybridized to the array using the miRNA complete labeling and hyb kit (Agilent, Santa Clara, CA, USA) and spike-in kit (Agilent). Data were generated on a GeneChip Scanner 3000 (Agilent), Human microRNA Microarray Release 16.0, 8x60K arrays (Agilent). Data were imported into R, RMA normalized using the package AgiMicroRna,^{46 (link)} summarized to the log2 scale and returned for further association analysis as a gene by sample matrix. The miR expression data are MIAME compliant and have been submitted to the Gene Expression Omnibus (GSE73774). Differences in microRNA expression were determined with Omics Explorer (Qlucore, Lund, Sweden), using ANOVA on 549 miRs from TCGA^{12 (link)} and 1368 miRs from cell lines using using a cutoff of P<0.05 and false discovery rate of 0.09 and 0.73, respectively, calculated using the Benjamini–Hochberg method. Fold change of miRs found to be significant (P<0.05) in both TCGA and cell lines was then determined by comparing expression in each cluster with the other two clusters (Table 3).