Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Comparative Study
, 8 Suppl 1 (Suppl 1), S5

Cross Platform Microarray Analysis for Robust Identification of Differentially Expressed Genes

Affiliations
Comparative Study

Cross Platform Microarray Analysis for Robust Identification of Differentially Expressed Genes

Roberta Bosotti et al. BMC Bioinformatics.

Abstract

Background: Microarrays have been widely used for the analysis of gene expression and several commercial platforms are available. The combined use of multiple platforms can overcome the inherent biases of each approach, and may represent an alternative that is complementary to RT-PCR for identification of the more robust changes in gene expression profiles. In this paper, we combined statistical and functional analysis for the cross platform validation of two oligonucleotide-based technologies, Affymetrix (AFFX) and Applied Biosystems (ABI), and for the identification of differentially expressed genes.

Results: In this study, we analysed differentially expressed genes after treatment of an ovarian carcinoma cell line with a cell cycle inhibitor. Treated versus control RNA was analysed for expression of 16425 genes represented on both platforms. We assessed reproducibility between replicates for each platform using CAT plots, and we found it high for both, with better scores for AFFX. We then applied integrative correlation analysis to assess reproducibility of gene expression patterns across studies, bypassing the need for normalizing expression measurements across platforms. We identified 930 genes as differentially expressed on AFFX and 908 on ABI, with approximately 80% common to both platforms. Despite the different absolute values, the range of intensities of the differentially expressed genes detected by each platform was similar. ABI showed a slightly higher dynamic range in FC values, which might be associated with its detection system. 62/66 genes identified as differentially expressed by Microarray were confirmed by RT-PCR.

Conclusion: In this study we present a cross-platform validation of two oligonucleotide-based technologies, AFFX and ABI. We found good reproducibility between replicates, and showed that both platforms can be used to select differentially expressed genes with substantial agreement. Pathway analysis of the affected functions identified themes well in agreement with those expected for a cell cycle inhibitor, suggesting that this procedure is appropriate to facilitate the identification of biologically relevant signatures associated with compound treatment. The high rate of confirmation found for both common and platform-specific genes suggests that the combination of platforms may overcome biases related to probe design and technical features, thereby accelerating the identification of trustworthy differentially expressed genes.

Figures

Figure 1
Figure 1
FACS analysis. A2780 cells were left untreated (NT) or treated for 6 hours with a cell cycle inhibitor (TRT). BrdU was added 30 min before harvesting and samples were processed for cell cycle analysis and BrdU incorporation analysis.
Figure 2
Figure 2
Intraplatform reproducibility. The descriptive CAT (Correspondence At the Top) plots [9] were used to evaluate the array-to-array precision within each microarray platform for the three replicates. CAT Plots describe the proportion of genes in common between replicates as function of list size. To generate CAT Plots on treated samples we used the lists of genes ranked by |log2(fold change)| created by each of the treated samples (TRT) versus a reference, which is one of the control samples (CTRL), both for AFFX (A.1) and ABI (A.2). Similarly we calculated CAT Plots for CTRL samples using one of the TRTs as reference for AFFX (B.1) and ABI (B.2).
Figure 3
Figure 3
Microarray analysis Pipeline. Microarray analysis pipeline applied to AFFX and ABI data to identify lists of differentially expressed genes.
Figure 4
Figure 4
Interplatform agreement. The descriptive CAT (Correspondence At the Top) plots [9] were used to evaluate interplatform agreement. The Correspondence at the top was evaluated for AFFX and ABI using the full set of genes after the IQR filtering step (black line) or after the IC filtering step (red line). For each of the two platforms average |log2 (Fold Change)|, calculated between treatment vs. control group, was used to generate CAT Plots.
Figure 5
Figure 5
Genes found differentially expressed using AFFX and ABI. A) Set of 726 genes passing SAM statistical validation both in AFFX and ABI, B) Set of 182 genes passing SAM statistical validation only in ABI, C) Set of 204 genes passing SAM statistical validation only in AFFX. AFFX data are reported in red, ABI in black. The lack of statistical significance in one of the platforms (B, C) is also associated with a limited log2(Fold Change) variation. On the other hand, log2(Fold Change) variation for genes passing the statistical validation in both platforms (A) is quite similar, although ABI seems to have wider log2(Fold Change) dynamic range.

Similar articles

See all similar articles

Cited by 21 articles

See all "Cited by" articles

References

    1. Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP, Coller H, Loh ML, Dowining JR, Caligiuri MA, Bloomfield CD, Lander ES. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science. 1999;286:531–7. doi: 10.1126/science.286.5439.531. - DOI - PubMed
    1. Chang HY, Nuyten DSA, Sneddon JB, Hastie T, Tibshirani R, Sorlie T, Dai H, He YD, van't Veer LJ, Bertelink H, van de Rijn M, Brown PB, van de Vijver MJ. Robustness, scalability, and integration of a wound-response gene expression signature in predicting breast cancer survival. Proc Natl Acad Sci USA. 2005;102:3738–3743. doi: 10.1073/pnas.0409462102. - DOI - PMC - PubMed
    1. Bild AH, Yao G, Chang JT, Wang Q, Potti A, Chasse D, Joshi MB, Harpole D, Lancaster JM, Berchuck A, Olson JA, Jr, Marks JR, Dressman HK, West M, Nevins JR. Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature. 2006;439:353–357. doi: 10.1038/nature04296. - DOI - PubMed
    1. Koch WH. Technology platforms for pharmacogenomic diagnostic assays. Nat Rev Drug Discov. 2004;3:749–761. doi: 10.1038/nrd1496. - DOI - PubMed
    1. Lam LT, Pickeral OK, Peng AC, Rosenwald A, Hurt EM, Giltnane JM, Averett LM, Zhao H, Davis RE, Sathyamoorthy M, Wahl LM, Harris ED, Mikovitis JA, Monks AP, Hollingshead MG, Sausville EA, Staudt LM. Genome-scale measurement of mRNA turnover and the mechanism of action of the anti-cancer drug flavopiridol. Genome Biol. 2001;2:RESEARCH0041. doi: 10.1186/gb-2001-2-10-research0041. - DOI - PMC - PubMed

MeSH terms

LinkOut - more resources

Feedback