Microarray analysis after RNA amplification can detect pronounced differences in gene expression using limma.

Diboun I, Wernisch L, Orengo CA, Koltzenburg M. · 2006

Read Microarray analysis after RNA amplification can detect pronounced differences in gene expression using limma. by Diboun I, Wernisch L, Orengo CA, Koltzenburg M. (2006) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc1618401.txt.

Introduction

Copying RNA to boost tiny samples can bend the very ratios labs want to measure. Diboun and co-authors quantified that risk in 2006 for a common amplification protocol. Boosting is unavoidable when tissue is scarce.

Distortion struck when a signal shift in one sample fell outside the scanner's range. The bend usually shrank ratios, weakening their statistical proof. Strong changes survived well: 87 percent of top genes kept their significance. Moderate changes kept it only 69 percent of the time. Diboun found the limma analysis method coped better with amplification than a rival approach.

Catalog record

CREATORDiboun I, Wernisch L, Orengo CA, Koltzenburg M.
YEAR2006
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC1618401

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Microarray analysis after RNA amplification can detect pronounced differences in gene expression using limma By Ilhem Diboun, Lorenz Wernisch, Christine Anne Orengo, Martin Koltzenburg Published in: BMC Genomics DOI: 10.1186/1471-2164-7-252 ABSTRACT Background RNA amplification is necessary for profiling gene expression from small tissue samples. Previous studies have shown that the T7 based amplification techniques are reproducible but may distort the true abundance of targets. However, the consequences of such distortions on the ability to detect biological variation in expression have not been explored sufficiently to define the true extent of usability and limitations of such amplification techniques. Results We show that expression ratios are occasionally distorted by amplification using the Affymetrix small sample protocol version 2 due to a disproportional shift in intensity across biological samples. This occurs when a shift in one sample cannot be reflected in the other sample because the intensity would lie outside the dynamic range of the scanner. Interestingly, such distortions most commonly result in smaller ratios with the consequence of reducing the statistical significance of the ratios. This becomes more critical for less pronounced ratios where the evidence for differential expression is not strong. Indeed, statistical analysis by limma suggests that up to 87% of the genes with the largest and therefore most significant ratios (p < 10e-20) in the unamplified group have a p-value below 10e-20 in the amplified group. On the other hand, only 69% of the more moderate ratios (10e-20 < p < 10e-10) in the unamplified group have a p-value below 10e-10 in the amplified group. Our analysis also suggests that, overall, limma shows better overlap of genes found to be significant in the amplified and unamplified groups than the Z-scores statistics. Conclusion We conclude that microarray analysis of amplified samples performs best at detecting differences in gene expression, when these are large and when limma statistics are used. BACKGROUND Microarray technology offers a high throughput approach to transcriptional profiling on a genome wide scale. However, the relatively large amount of starting material required for standard hybridization has limited its full potential. In complex biological systems such as the nervous system, the utility of this approach is complicated by the fact that even in anatomically discrete regions, many divergent cell types are intermingled. It is often desirable to investigate gene expression profiles of distinct cell types and although laser microdissection provides a solution to the problem of tissue procurement, the small amount of RNA that can be harvested has precluded a straightforward combination of both technologies. This limitation is compounded by the need of replication essential for statistical analysis. Another scenario in which the lack of sufficient tissue availability has been challenging is the correlation of the phenotype in individual experimental animals with comprehensive gene expression profiles. For example, so far it has not been possible to correlate the inter-individual behavioural variability in animal models of chronic pain with the corresponding correlates in gene expression in the principal anatomical components of the pain pathway as such structures in individual animals do not yield sufficient amounts of RNA for standard hybridization protocols. To overcome these issues, increasingly sophisticated approaches to RNA amplification from small tissue samples have been developed for use with microarrays. These fall principally into two categories. One is based on PCR and is characterized by an exponential increase in copy number while the other is based on the T7-polymerase in-vitro transcription (IVT) to achieve a linear amplification of targets. For maximum fidelity, linearity of target amplification is desirable. Thus, substantial work has focused on exploring the T7 linear techniques [1-4], in particular the Affymetrix small sample protocol II has been assessed by numerous studies with interesting results. Analysis based on correlating intensity levels and assessing concordance in detection calls has indicated a high level of reproducibility [2,3,5,6]. However, occasional failure to maintain the true abundance level of transcripts was also found [3,5,6] due to the protocol 3'bias [1,2,4,6,7]. Such bias is thought to be related to the use of random hexamers to prime the RT reaction in the additional round of amplification. With priming that is remote from the 3'end, RT may not be successfully completed causing a diminution in the signal from the 5' regions. These observations are useful indicators of protocol validity; though, the ultimate fidelity criterion is the ability to maintain differential expression between different tissues or under varying experimental conditions. Some previous studies have reported a 50% drop in significant changes in gene expression using RNA amplification [4], suggesting that RNA amplification may suffer major problems and is potentially unsuitable for microarray analysis. In this study, we critically appraise the suitability and merits of transcript amplification from small biological samples for microarray expression profiling using the Affymetrix small sample protocol II. We confirm previous findings on the reproducibility and fidelity of the protocol. We present compelling evidence for the 3' bias introduced by RNA amplification. We show how distortions in intensity may be reflected in the expression ratios from biologically different samples. Importantly, we explore the effect of distortions in expression ratios on their statistical significance. RESULTS In this study, we undertake a detailed analysis of target amplification for microarrays using the Affymetrix small sample protocol II. This analysis was performed using control data from standard protocol preparations as reference. For the rest of the article, we refer to the standard protocol as the OneRA (one round amplification) protocol and the small sample protocol as the TwoRA (two rounds amplification) protocol. TwoRA and OneRA samples from three different biological groups were used in this study: the DRG (dorsal root ganglia) group, the SN (spinal cord from normal animals) group and the SA (spinal cord from animals with axotomised sciatic nerve) group. In each group, there are 4 replicates from the TwoRA and 3 replicates from the OneRA (Figure 1a). While, the main objective of this study is to assess the extent to which biologically relevant variation in gene expression can be detected in the TwoRA, we begin by assessing the reproducibility of the TwoRA protocol and its fidelity in maintaining the relative abundance of targets. Reproducibility and fidelity in maintaining expression levels Scatter plots of log2 intensities from paired TwoRA replicates show expectedly high level of consistency similar to that observed with the OneRA replicates (Figure 1b &1c) ((r) ranging from 0.990 to 0.994 among all possible pairs of the TwoRA replicates and the OneRA replicates from all biological groups). However, comparing the average log2 intensity values from the OneRA and the TwoRA (Figure 1d) for a single tissue, we see evidence of variability implying that the TwoRA protocol may distort the signal. We used an ANOVA approach to confirm that the variability between protocol groups is greater than that among replicates within each group. In particular, a one-way, two-levels ANOVA analysis was performed for each gene separately, with 3 measurements from the OneRA (level 1) and 4 measurements from the TwoRA (level 2). First, the between group mean sum of squares MSA as well as the mean residual sum of squares MSE were calculated. The median of the MSA was higher than the median of the MSE (given in parenthesis) in all biological groups: DRG 0.050 (0.023), SN 0.062 (0.016), SA 0.068. To

Read the complete text