The molecular portraits of breast tumors are conserved across microarray platforms

Abstract

Background: Validation of a novel gene expression signature in independent data sets is a critical step in the development of a clinically useful test for cancer patient risk-stratification. However, validation is often unconvincing because the size of the test set is typically small. To overcome this problem we used publicly available breast cancer gene expression data sets and a novel approach

A high-resolution integrated analysis of genetic and expression profiles of breast cancer cell lines

“…The Hu et al [3] dataset was used as a training set to predict basal and luminal phenotypes using the prediction analysis of microarrays package (PAM), as previously described [3,20]. Briefly, 24,000 probes from Hu et al [3] were filtered to remove probes flagged as absent in more than 10% of the samples.…”

“…Briefly, 24,000 probes from Hu et al [3] were filtered to remove probes flagged as absent in more than 10% of the samples. Probes with the greatest variation across samples were then selected (those with an inter-quartile range of log ratio values of more than 0.65).…”

“…This left 4,005 genes for further analysis. The Hu et al [3] dataset was combined with the 17 K expression dataset on the basis of Unigene accession number. Expression data were then centred and missing values were imputed using a k nearest neighbour algorithm.…”

“…A PAM predictor of genes intersecting with cell line expression profiles was constructed which correctly classified 100% of luminal and basal tumours in Hu et al [3].…”

“…Microarray-based RNA expression profiling has been extensively used as a tool to help unravel the complexity of breast cancer [1][2][3][4][5][6]. These expression profiling studies have contributed to a working model of breast tumour taxonomy that suggests the existence of at least four distinct subgroups that have both biological and clinical significance.…”