A comparison of normalization techniques for microRNA microarray data

Normalization of expression levels applied to microarray data can help in reducing measurement error. Different methods, including cyclic loess, quantile normalization and median or mean normalization, have been utilized to normalize microarray data. Although there is considerable literature regarding normalization techniques for mRNA microarray data, there are no publications comparing normalization techniques for microRNA (miRNA) microarray data, which are subject to similar sources of measurement error. In this paper, we compare the performance of cyclic loess, quantile normalization, median normalization and no normalization for a single-color microRNA microarray dataset. We show that the quantile normalization method works best in reducing differences in miRNA expression values for replicate tissue samples. By showing that the total mean squared error are lowest across almost all 36 investigated tissue samples, we are assured that the bias correction provided by quantile normalization is not outweighed by additional error variance that can arise from a more complex normalization method. Furthermore, we show that quantile normalization does not achieve these results by compression of scale.     

Identification of Most Stable Endogenous Control Genes for MicroRNA Quantification in the Developing Mouse Lung

MicroRNAs (miRNAs) are endogenous small non coding RNAs acting as negative regulators. miRNA are involved in lung development and pulmonary diseases. Measurement of their levels by qPCR is directly influenced by the stability of normalization gene(s), which can be affected by the experimental conditions. The developing lung is a changing tissue and one normalization gene showing stability on one developmental day may be modulated over time. Moreover, some developmental events are affected by sex, which also has to be considered. In this study, we compared stability of five putative control genes in the lung between sexes from the pseudoglandular to the alveolar stages and in adult lungs. Expression of sno135, sno142, sno202, sno234, and sno251 was studied by qPCR in male and female lung samples collected at seven time points from GD 15.5 to PN 30. Cq values of sno251 showed the highest variation across the different developmental stages, while sno234 was the most stable gene. Gene expression stability was studied by geNorm, NormFinder and BestKeeper. Our data showed that ranking of genes based on expression stability changed according to developmental time and sex. sno135/sno234 and sno142/sno234 were proposed as best combinations of normalization genes when both sexes and all the studied developmental stages are considered. Normalization of let7-a RNA levels with different pairs of control genes proposed by geNorm and NormFinder gave similar data, while the use of less stable genes introduced a statistically significant difference on PN 0. In conclusion, variations in stability of normalization gene expression are observed over time and according to sex during lung development. Best pairs of normalization genes are presented for specific developmental stages, and for the period extending from the pseudoglandular to the alveolar stages. The use of normalization genes selected for their expression stability is essential in lung development studies.     

Identification of stably expressed reference small non‐coding RNAs for microRNA quantification in high‐grade serous ovarian carcinoma tissues

MicroRNAs (miRNAs) belong to a family of small non‐coding RNAs (sncRNAs) playing important roles in human carcinogenesis. Multiple investigations reported miRNAs aberrantly expressed in several cancers, including high‐grade serous ovarian carcinoma (HGS‐OvCa). Quantitative PCR is widely used in studies investigating miRNA expression and the identification of reliable endogenous controls is crucial for proper data normalization. In this study, we aimed to experimentally identify the most stable reference sncRNAs for normalization of miRNA qPCR expression data in HGS‐OvCa. Eleven putative reference sncRNAs for normalization (U6, SNORD48, miR‐92a‐3p, let‐7a‐5p, SNORD61, SNORD72, SNORD68, miR‐103a‐3p, miR‐423‐3p, miR‐191‐5p, miR‐16‐5p) were analysed on a total of 75 HGS‐OvCa and 30 normal tissues, using a highly specific qPCR. Both the normal tissues considered to initiate HGS‐OvCa malignant transformation, namely ovary and fallopian tube epithelia, were included in our study. Stability of candidate endogenous controls was evaluated using an equivalence test and validated by geNorm and NormFinder algorithms. Combining results from the three different statistical approaches, SNORD48 emerged as stably and equivalently expressed between malignant and normal tissues. Among malignant samples, considering groups based on residual tumour, miR‐191‐5p was identified as the most equivalent sncRNA. On the basis of our results, we support the use of SNORD48 as best reference sncRNA for relative quantification in miRNA expression studies between HGS‐OvCa and normal controls, including the first time both the normal tissues supposed to be HGS‐OvCa progenitors. In addition, we recommend miR‐191‐5p as best reference sncRNA in miRNA expression studies with prognostic intent on HGS‐OvCa tissues.     

Effect of irradiation on the expression of DNA repair genes studied in human fibroblasts by real-time qPCR using three methods of reference gene validation

The aim of this study was to determine the effects of ionizing radiation on gene expression by using for a first time a qPCR platform specifically established for the detection of 94 DNA repair genes but also to test the robustness of these results by using three analytical methods (global pattern recognition, ΔΔCq/Normfinder and ΔΔCq/Genorm). Study was focused on these genes because DNA repair is known primarily to determine the radiation response. Six strains of normal human fibroblasts were exposed to 2 Gy, and changes in gene expression were analyzed 24 h thereafter. A significant change in gene expression was found for only few genes, but the genes detected were mostly different for the three analytical methods used. For GPR, a significant change was found for four genes, in contrast to the eight or nine genes when applying ΔΔCq/Genorm or ΔΔCq/Normfinder, respectively. When using all three methods, a significant change in expression was only seen for GADD45A and PCNA. These data demonstrate that (1) the genes identified to show an altered expression upon irradiation strongly depend on the analytical method applied, and that (2) overall GADD45A and PCNA appear to play a central role in this response, while no significant change is induced for any of the other DNA repair genes tested.     

Identification of a 4-microRNA signature for clear cell renal cell carcinoma metastasis and prognosis

Renal cell carcinoma (RCC) metastasis portends a poor prognosis and cannot be reliably predicted. Early determination of the metastatic potential of RCC may help guide proper treatment. We analyzed microRNA (miRNA) expression in clear cell RCC (ccRCC) for the purpose of developing a miRNA expression signature to determine the risk of metastasis and prognosis. We used the microarray technology to profile miRNA expression of 78 benign kidney and ccRCC samples. Using 28 localized and metastatic ccRCC specimens as the training cohort and the univariate logistic regression and risk score methods, we developed a miRNA signature model in which the expression levels of miR-10b, miR-139-5p, miR-130b and miR-199b-5p were used to determine the status of ccRCC metastasis. We validated the signature in an independent 40-sample testing cohort of different stages of primary ccRCCs using the microarray data. Within the testing cohort patients who had at least 5 years follow-up if no metastasis developed, the signature showed a high sensitivity and specificity. The risk status was proven to be associated with the cancer-specific survival. Using the most stably expressed miRNA among benign and tumorous kidney tissue as the internal reference for normalization, we successfully converted his signature to be a quantitative PCR (qPCR)-based assay, which showed the same high sensitivity and specificity. The 4-miRNA is associated with ccRCC metastasis and prognosis. The signature is ready for and will benefit from further large clinical cohort validation and has the potential for clinical application.     

Comparison of normalization methods with microRNA microarray

MicroRNAs (miRNAs) are a group of RNAs that play important roles in regulating gene expression and protein translation. In a previous study, we established an oligonucleotide microarray platform to detect miRNA expression. Because it contained only hundreds of probes, data normalization was difficult. In this study, the microarray data for eight miRNAs extracted from inflamed rat dorsal root ganglion (DRG) tissue were normalized using 15 methods and compared with the results of real-time polymerase chain reaction. It was found that the miRNA microarray data normalized by the print-tip loess method were the most consistent with results from real-time polymerase chain reaction. Moreover, the same pattern was also observed in 14 different types of rat tissue. This study compares a variety of normalization methods and will be helpful in the preprocessing of miRNA microarray data.     

Identification of reference microRNAs and suitability of archived hemopoietic samples for robust microRNA expression profiling

In many cancers, including neuroblastoma, microRNA (miRNA) expression profiling of peripheral blood (PB) and bone marrow (BM) may increase understanding of the metastatic process and lead to the identification of clinically informative biomarkers. The quality of miRNAs in PB and BM samples archived in PAXgene? blood RNA tubes from large-scale clinical studies and the identity of reference miRNAs for standard reporting of data are to date unknown. In this study, we evaluated the reliability of expression profiling of 377 miRNAs using quantitative polymerase chain reaction (qPCR) in PB and BM samples (n = 90) stored at ?80 °C for up to 5 years in PAXgene? blood RNA tubes. There was no correlation with storage time and variation of expression for any single miRNA (r < 0.50). The profile of miRNAs isolated as small RNAs or co-isolated with small/large RNAs was highly correlated (r = 0.96). The mean expression of all miRNAs and the geNorm program identified miR-26a, miR-28-5p, and miR-24 as the most stable reference miRNAs. This study describes detailed methodologies for reliable miRNA isolation and profiling of PB and BM, including reference miRNAs for qPCR normalization, and demonstrates the suitability of clinical samples archived at ?80 °C into PAXgene? blood RNA tubes for miRNA expression studies.     

Reliable gene signatures for microarray classification: assessment of stability and performance

MOTIVATION: Two important questions for the analysis of gene expression measurements from different sample classes are (1) how to classify samples and (2) how to identify meaningful gene signatures (ranked gene lists) exhibiting the differences between classes and sample subsets. Solutions to both questions have immediate biological and biomedical applications. To achieve optimal classification performance, a suitable combination of classifier and gene selection method needs to be specifically selected for a given dataset. The selected gene signatures can be unstable and the resulting classification accuracy unreliable, particularly when considering different subsets of samples. Both unstable gene signatures and overestimated classification accuracy can impair biological conclusions. METHODS: We address these two issues by repeatedly evaluating the classification performance of all models, i.e. pairwise combinations of various gene selection and classification methods, for random subsets of arrays (sampling). A model score is used to select the most appropriate model for the given dataset. Consensus gene signatures are constructed by extracting those genes frequently selected over many samplings. Sampling additionally permits measurement of the stability of the classification performance for each model, which serves as a measure of model reliability. RESULTS: We analyzed a large gene expression dataset with 78 measurements of four different cartilage sample classes. Classifiers trained on subsets of measurements frequently produce models with highly variable performance. Our approach provides reliable classification performance estimates via sampling. In addition to reliable classification performance, we determined stable consensus signatures (i.e. gene lists) for sample classes. Manual literature screening showed that these genes are highly relevant to our gene expression experiment with osteoarthritic cartilage. We compared our approach to others based on a publicly available dataset on breast cancer. AVAILABILITY: R package at http://www.bio.ifi.lmu.de/~davis/edaprakt     

A systematic study of normalization methods for Infinium 450K methylation data using whole-genome bisulfite sequencing data

DNA methylation plays an important role in disease etiology. The Illumina Infinium HumanMethylation450 (450K) BeadChip is a widely used platform in large-scale epidemiologic studies. This platform can efficiently and simultaneously measure methylation levels at ∼480,000 CpG sites in the human genome in multiple study samples. Due to the intrinsic chip design of 2 types of chemistry probes, data normalization or preprocessing is a critical step to consider before data analysis. To date, numerous methods and pipelines have been developed for this purpose, and some studies have been conducted to evaluate different methods. However, validation studies have often been limited to a small number of CpG sites to reduce the variability in technical replicates. In this study, we measured methylation on a set of samples using both whole-genome bisulfite sequencing (WGBS) and 450K chips. We used WGBS data as a gold standard of true methylation states in cells to compare the performances of 8 normalization methods for 450K data on a genome-wide scale. Analyses on our dataset indicate that the most effective methods are peak-based correction (PBC) and quantile normalization plus β-mixture quantile normalization (QN.BMIQ). To our knowledge, this is the first study to systematically compare existing normalization methods for Illumina 450K data using novel WGBS data. Our results provide a benchmark reference for the analysis of DNA methylation chip data, particularly in white blood cells.     

Suitable internal control microRNA genes for measuring miRNA abundance in pig milk during different lactation periods

Determination of an optimal set/number of internal control microRNA (miRNA) genes is a critical, but often undervalued, detail of quantitative gene expression analysis. No validated internal genes for miRNA quantitative PCR (q-PCR) in pig milk were available. We compared the expression stability of six porcine internal control miRNA genes in pig milk from different lactation periods (1 h, 3 days, 7 days, 14 days, 21 days, and 28 days postpartum), using an EvaGreen q-PCR approach. We found that using the three most stable internal control genes to calculate the normalization factor is sufficient for producing reliable q-PCR expression data. We also found that miRNAs are superior to ribosomal RNA (rRNA) and snRNA, which are commonly used as internal controls for normalizing miRNA q-PCR data. In terms of economic and experimental feasibility, we recommend the use of the three most stable internal control miRNA genes (miR-17, -107 and -103) for calculating the normalization factors for pig milk samples from different lactation periods. These results can be applied to future studies aimed at measuring miRNA abundance in porcine milk.     

Representation Analysis of miRNA in Urine Microvesicles and Cell-Free Urine in Prostate Diseases

Urine of prostate cancer patients contains tumor-specific biopolymers, including protein- and microvesicles-associated miRNAs that can potentially be used as oncomarkers. Previously we have characterized urinary extracellular vesicles and demonstrated diagnostic potential of their miRNAs. In this study, we have performed a comparative analysis of 84 miRNA in paired samples of urine microvesicles and urine supernatant from healthy men, patients with benign prostate hyperplasia, and prostate cancer patients using miRCURY LNA miRNA qPCR Panels. In all groups of patients, miRNA subsets characterized by different distribution between the urinary fractions have been found. In this context, two groups of miRNAs have been identified, which are involved in several signaling pathways including those associated with prostate cancer development.