Comparison between MPSS and Illumina bead array results
We have previously used EST scan and MPSS to analyze pooled samples of ESCs and EBs from three different WiCell lines (H1, H7 and H9) [5]. Comparison between the two methodologies indicated that while there is good concordance for genes expressed at high levels, this does not hold for genes expressed at lower levels. As a test of the quality of the data generated in these experiments and to evaluate whether comparisons can be made across different methodologies, we re-ran the identical samples on the bead array platform. The complete comparison of gene expression is shown in Additional file 2 and is summarized in Tables 3 and Table 4. Overall, concordance in Illumina array was better than that evident between EST scan and MPSS datasets [9], but clearly showed much wider differences than that seen with running duplicates in the same assay format. Nevertheless, this comparison provides an independent verification of the data and suggests that if a sample is detected in more than one large-scale analysis, the reliability of the gene expression detection is high, which also reduces the number of individual genes needed to be verified. Caution should be observed in comparing different samples run on different platforms, especially when there has not been rigorous bioinformatic matching of the source sequences used to identify genes in the platforms. Often genes called by the same symbol originate from different database records, which may originate from different splice variants or contain sequence differences due to polymorphisms or outright error [19].
Table 3  Expression of hESC specific markers in pooled hESC sample as detected by Illumina bead array The expression of previously identified hESC markers was examined in all hESC samples (the values displayed represent the expression level of pooled H1, H7 and H9). Most of the genes were also identified using Illumina bead array in all 8 hESC populations in this study (1*), the gene CER1 was detected in all except one duplicate of H9 (2*), Nanog was not detected in all populations (3*) and Sox2, Lin41, NR6A1 and FoxD3 were not detected in the array although they were present in the chips for hybridization (4*).
Table 4  Comparison of MPSS and Illumina bead array results The samples were analyzed by MPSS and bead array. The number of genes detected by each method and the degree of overlap is summarized. Note much higher degree of overlap when the top 2000 hits were compared. *: Most of the genes detected by MPSS were novel genes not included in the bead array.

H