Reproducibility of gene expression across generations of Affymetrix microarrays

Abstract

Background: The development of large-scale gene expression profiling technologies is rapidly changing the norms of biological investigation. But the rapid pace of change itself presents challenges. Commercial microarrays are regularly modified to incorporate new genes and improved target sequences. Although the ability to compare datasets across generations is crucial for any long-term research project, to date no means to allow such comparisons have been developed. In this study the reproducibility of gene expression levels across two generations of Affymetrix GeneChips (HuGeneFL and HG-U95A) was measured.

Results: Correlation coefficients were computed for gene expression values across chip generations based on different measures of similarity. Comparing the absolute calls assigned to the individual probe sets across the generations found them to be largely unchanged.

Conclusion: We show that experimental replicates are highly reproducible, but that reproducibility across generations depends on the degree of similarity of the probe sets and the expression level of the corresponding transcript.

Figure 1

Distribution of the number of probe sets versus the number of probe pairs in common is shown for the 8,044 probe sets on both the older HuGeneFL and the newer HG-U95A Chips. The Y-axis of the line graph appears on the left, while the Y-axis of the column graph appears on the right. Each of the 8,044 probe sets is represented in only one column. All expression measurements across all 14 chips were used to calculate each column's correlation coefficient. The effect of the number of common probe pairs on the correlation of the gene expression values is illustrated. The correlation between the gene expression values across the two generations of chips increases as the number of common probe pairs increases; however, there are few probe sets with many common probe pairs.

Figure 2

Distribution of the number of probe sets versus the number of 'P' calls given by those probe sets for all the 14 chips. The Y-axis of the line graph appears on the left, while the Y-axis of the column graph appears on the right. The correlation between the gene expression values of HuGeneFL and HG-U95A is roughly correlated to the number of 'P' calls given by the probe sets in each chip.

Figure 3

Distribution of the number of probe sets versus the different levels of gene expression values on the HuGeneFL. The Y-axis of the line graph appears on the left, while the Y-axis of the column graph appears on the right. The correlation between the gene expression values from the HuGeneFL and HG-U95A is roughly correlated to the levels of gene expression levels on the HuGeneFL chip. For the Y-axis, each of the 8,044 probe sets was represented 7 times, one for each measurement on the HuGeneFL chip.

Figure 4

Distribution of 8,044 correlation coefficients corresponding to the common probe sets between HuGeneFL and HG-U95A. Representative probe sets are illustrated in the smaller plots (Affymetrix probe set ID numbers: M14539_at, 38052_at, X16504_s_at and 2035_s_at). The lower left graph shows the plot for a probe set whose expression values were positively correlated between the two chips, while lower right graph shows the plot for a probe set whose expression values were negatively correlated.

Figure 5

Intensity plots of HuGeneFL vs. HG-U95A for different number of common probe pairs (0, 1, 8 and 14) for the probe sets across the two generations. Each of the probe sets was represented 7 times on X-axis and Y-axis, one for each measurement on the HuGeneFL and HG-U95A chip respectively.

Figure 6

Graph showing the correlation between the gene expression of HuGeneFL and HG-U95A for different levels of expression on HuGeneFL chip. High expression levels appear to compensate for low numbers of common probe pairs between chip generations.