Comparison of hybridization-based and sequencing-based gene expression technologies on biological replicates

Abstract

Background: High-throughput systems for gene expression profiling have been developed and have matured rapidly through the past decade. Broadly, these can be divided into two categories: hybridization-based and sequencing-based approaches. With data from different technologies being accumulated, concerns and challenges are raised about the level of agreement across technologies. As part of an ongoing large-scale cross-platform data comparison framework, we report here a comparison based on identical samples between one-dye DNA microarray platforms and MPSS (Massively Parallel Signature Sequencing).

Results: The DNA microarray platforms generally provided highly correlated data, while moderate correlations between microarrays and MPSS were obtained. Disagreements between the two types of technologies can be attributed to limitations inherent to both technologies. The variation found between pooled biological replicates underlines the importance of exercising caution in identification of differential expression, especially for the purposes of biomarker discovery.

Conclusion: Based on different principles, hybridization-based and sequencing-based technologies should be considered complementary to each other, rather than competitive alternatives for measuring gene expression, and currently, both are important tools for transcriptome profiling.

Figure 1

Correspondence of expression detection between microarrays and MPSS. Chi-square statistics was used as a measure of data correspondence in terms of absent/present call made by each microarray platform versus each MPSS library. For microarray data, a fixed definition of the "Present/Absent" call based on quality flags and filtering status for each gene. For MPSS data, a gene (UniGene clusters) was considered "Present" if the copy number was above a given threshold, otherwise it was considered "Absent". The "Present/Absent" call threshold varied from 1 to 100 tpm for MPSS in steps of 1 and Chi-square statistics were calculated for each threshold. Part (a) shows how the p-values from these comparisons varied with threshold on data for MRP1, and part (b) shows the same for MRP2.

Figure 2

Correspondence of expression level between microarrays and MPSS. Pearson correlation was used as a measure of correspondence of expression levels reported by each microarray platform versus each MPSS library. For microarray data, normalized and filtered intensities were used. With the tag count threshold being varied from 1 to 100 tpm for MPSS in steps of 1, Pearson correlations were calculated for the genes with tag counts above each given threshold. Part (a) shows how the correlations from these comparisons varied with threshold on data for MRP1, and part (b) shows the same for MRP2.

Figure 3

CAT plots using MPSS data as reference. Correspondence between MPSS and microarray platforms on detection of highly expressed genes was assessed by CAT plots. In Figure 3, the x-axis represents the sizes of gene sets at the top of gene expressions, from 10 to 800 at a step of 5, and the y-axis shows the percentage of overlapping genes between each microarray platform and MPSS in the given set of genes. Part (a) represents MRP1, and part (b) shows MRP2.

Figure 4

Statistics of differentially expressed genes identified by SAM as a function of "delta" threshold. In Figure 4, the x-axis represents the threshold ("delta") for differential expression identification which changes from 0.1 to 1.2 at a step of 0.1, while the y-axis is in (a) the percentage of genes which were considered to be differentially expressed at a given threshold; and in (b) false discovery rate.