Introduction

The GeneChip Human Gene 1.0 ST array (Affymetrix, Santa Clara, CA, USA) is a whole transcript–based array for gene expression profiling. It was designed using a subset of the probes from the Human Exon 1.0 ST array (Affymetrix) and includes only well-annotated exons based on RefSeq (www.ncbi.nlm.nih.gov/refseq), Ensembl (www.ensembl.org), and putative complete coding sequences from GenBank (www.ncbi.nlm.nih.gov/genbank). An important feature of the array is that, as for the Human Exon 1.0 ST array, it queries the entire transcript in contrast to most older Affymetrix arrays that query the 3′ end of transcripts. Another feature of this array is the use of a target labeling protocol that generates biotinylated sense DNA instead of biotinylated cRNA. As a result, the target-probe hybridization generates DNA-DNA duplexes, which are more specific than the RNA-DNA duplexes generated in hybridizations using the standard protocol for 3′-based expression array (1). Affymetrix has compared the performance of the Human Gene 1.0 ST array with the performance of the Human Exon 1.0 ST and Human Genome (HG) U133 Plus 2.0 arrays (Affymetrix), which have previously been shown to be highly correspondent (2). They observed a slightly better reproducibility for the Gene 1.0 ST array and comparable detection thresholds (www.affymetrix.com/support/technical/whitepapers/hugene_perf_whitepaper.pdf). Thus, the Gene 1.0 ST array offers an acceptable alternative to traditional 3′-based expression arrays.

The Illumina BeadChip array (San Diego, CA, USA) is another widely used platform for gene expression profiling (3,–4). In contrast to Affymetrix arrays that use 25-mer oligonucleotide probes, this array uses 50 bp oligonucleotide probes. A strong design preference is given to probes from sequences close to the 3′ end of the transcript, although there is no set minimum distance. Despite their physical differences, the Illumina HumanRef-8 BeadChip and the Affymetrix HG-U133 Plus 2.0 arrays exhibit very high agreement in terms of differentially expressed genes measured between two tissues (5,–6). Therefore, it would be interesting to determine if a similar degree of agreement is found between Illumina HumanRef-8 BeadChip and the Affymetrix Human Gene 1.0 ST array in order to provide evidence of the reliability of these platforms.

We have performed an analysis of the reproducibility and differential gene expression call concordance of Illumina HumanRef-8 BeadChip (referred to as Illumina BeadChip), Affymetrix Human Gene 1.0 ST, and Affymetrix HG-U133 Plus 2.0 arrays using the same RNA samples as were used in the MicroArray Quality Control Project (MAQC) (6). Additionally, we have compared the performance of the HG-U133 Plus 2.0 array using two different target labeling methods: one that generates biotinylated cRNA and requires 100 ng of total RNA (Message Amp from Ambion, Austin, TX, USA) and another that generates biotinylated cDNA and requires as little as 1 ng of total RNA (Ovation from NuGen, San Carlos, CA, USA) (7). Overall, we have four different technical protocols: HG-U133 with cRNA targets (referred to hereafter as Protocol A), HG-U133 with cDNA targets (Protocol N), Gene 1.0 ST (Protocol G) and Illumina BeadChip (Protocol I). It should be noted that although the protocols differ in many aspects, both the NuGen Ovation and the Affymetrix Gene 1.0 ST labeling methods produce biotinylated DNA hybridization targets whereas both Ambion labeling protocols used for the Affymetrix 3′ expression arrays and the Illumina BeadChip arrays produce biotinylated RNA hybridization targets. This set of experiments allows one to rank the impact of the array design as well as hybridization target on the concordance of the results.

Materials and Methods

RNA Samples and Experimental Design

The RNAs used in the MAQC Project (6)—Universal Human Reference RNA (Stratagene, La Jolla, CA, USA) and Human Brain Total RNA (Ambion)—were selected and referred to as Sample A and B, respectively. Sample labelings and hybridizations were performed on two different days, each time two technical replicates, for a total of four technical replicates per condition. RNA quality was first checked for chemical purity using a NanoDrop spectrophotomer (NanoDrop Technologies, Wilmington, DE, USA) and then assessed for RNA integrity using the Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). Both RNA types were of acceptable quality (RIN of 8.8 and 7.7 for Samples A and B, respectively).