2. Experimental Section
MicroRNA expression profiles were downloaded from Gene Expression Omnibus [11] GSE32935 (aging heart; [12] Exiqon Mouse microRNA v11.0) and GSE35672 (iPS-cardiomyocytes; [13], Illumina Human v2 MicroRNA expression beadchip datasets). microRNA microarray data from ventricular maturation (MirVana v2.0, miRBase version 8.0) were already available to us, as described by Chinchilla et al. [14]. To minimize technical (non-biological) variability among arrays, each data group was independently log2 transformed and then normalized using the quantiles normalization function implemented in the Bioconductor limma package [15], with default parameters run in R software [16], and finally, each probe was Z-scored (). MicroRNA expression data in ventricular development [14], also Z-scored, was identically processed as describe above, but with an initial k-nearest neighbor(KNN) imputation of the densitometry values <1 using the KNN algorithm implemented in the Bioconductor impute package [15] with default parameters. Hierarchical clustering (Euclidean distance and complete linkage), an unsupervised way of grouping samples based only on their gene expression similarities, was carried out using TM4 software suite [17].