Normalization and analysis of DNA microarray data by self-consistency and local regression

Our plain-English summary of Normalization and analysis of DNA microarray data by self-consistency and local regression, by T.B. Kepler, L. Crosby, K.T. Morgan (Genome Biology, 2002).

The study

• Authors: T.B. Kepler, L. Crosby, K.T. Morgan
• Journal: Genome Biology
• Published: 2002;3(7):RESEARCH0037
• Cited by: 64 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Background: With the advent of DNA hybridization microarrays comes the remarkable ability, in principle, to simultaneously monitor the expression levels of thousands of genes. The quantitative comparison of two or more microarrays can reveal, for example, the distinct patterns of gene expression that define different cellular phenotypes or the genes induced in the cellular response to insult or changing environmental conditions. Normalization of the measured intensities is a prerequisite of such comparisons, and indeed, of any statistical analysis, yet insufficient attention has been paid to its systematic study. The most straightforward normalization techniques in use rest on the implicit assumption of linear response between true expression level and output intensity. We find that these assumptions are not generally met, and that these simple methods can be improved.

Results: We have developed a robust semi-parametric normalization technique based on the assumption that the large majority of genes will not have their relative expression levels changed from one treatment group to the next, and on the assumption that departures of the response from linearity are small and slowly varying. We use local regression to estimate the normalized expression levels as well as the expression level-dependent error variance.

Conclusions: We illustrate the use of this technique in a comparison of the expression profiles of cultured rat mesothelioma cells under control and under treatment with potassium bromate, validated using quantitative PCR on a selected set of genes. We tested the method using data simulated under various error models and find that it performs well.

In plain English

This article described a statistical method for adjusting, or normalizing, DNA microarray data before comparing gene activity across arrays. It was not a study of asbestos exposure. The authors reported that simple normalization methods assume a straight-line link between true gene expression and measured signal, an assumption not generally met. Their method assumed most genes did not change between treatment groups and used a technique called local regression. They illustrated it with four microarrays of cultured rat mesothelioma cells, untreated and treated with potassium bromate. Two genes, Waf-1 and HSP-70, seemed to be switched up after treatment, and a selected set of genes was checked by quantitative PCR. The authors also tested the method on simulated data. In one simulated setting (10% of genes changed 1.5-fold, with distortion like that in the real data), the new method's power, the share of true changes detected, was 0.300, against 0.127 for a simple method. With added background noise in that setting, its false-positive score (average false detections divided by the number expected at an error rate of 0.01) was 0.888, against 7.791. The authors concluded that the method performs well.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: statistical methods paper: a new normalization technique, illustrated on a laboratory microarray experiment and tested on simulated data
• Who was studied: cultured rat mesothelioma cells, analysed on four Clontech microarrays (control and potassium bromate-treated)
• Exposure: potassium bromate treatment of cultured cells, compared with untreated controls; no asbestos exposure was studied
• Main results:
  • The authors found that the straight-line assumption behind simple normalization methods was not generally met, and that those methods can be improved
  • In the potassium bromate-treated rat mesothelioma cells, Waf-1 and HSP-70 seemed to be upregulated
  • In data simulated with the homoscedastic error model (Table 1), in the setting with 10% of genes changed, a treated-to-control ratio of 1.5 and bias as measured in the real data, power was 0.127 for the naive method and 0.300 for NoSeCoLoR. Power is the mean number of correct discriminations divided by the number of true changes.
  • In data simulated with multiplicative plus additive (background) error (Table 3), in the same setting, the false-positive score was 7.791 for the naive method and 0.888 for NoSeCoLoR. The score is the mean number of incorrect discriminations divided by the number expected at a nominal type-I error rate of 0.01.
• The authors' conclusion: The authors concluded that the method performed well when tested on data simulated under several error models

Topics

Medical terms in this study, explained in our Lexicon: Mesothelioma.

Related studies

• Analysis of the effect of occupational exposure to asbestos based on threshold regression modeling of case-control data — A. Chambaz, D. Choudat, et al. (2014)
• Ordinal regression models for epidemiologic data — B.G. Armstrong, M. Sloan (1989)
• Asbestos exposure and cancer mortality among petroleum refinery workers: a Poisson regression analysis of updated data — F. Montanaro, M. Ceppi, et al. (2004)
• Combined Analysis of SNP Array Data Identifies Novel CNV Candidates and Pathways in Ependymoma and Mesothelioma — G. Wajnberg, B.S. Carvalho, et al. (2015)
• Complete regression of human malignant mesothelioma xenografts following local injection of midkine promoter-driven oncolytic adenovirus — S. Kubo, Y. Kawasaki, et al. (2010)
• Relation of occupational exposure to respiratory symptoms and asthma in a general population sample: self-reported versus interview-based exposure data — P.S. Bakke, R. Hanoa, et al. (2001)

Browse return to medical research.