| ECCB 2002 Poster sorted by: Author | Number Next | Previous poster (in order of the view you have selected) |
Title: A 2nd Order Polynomial Normalization for Competitive Microarray Experiments | P85 |
Kroll, Torsten; Wölfl, Stefan torsten.kroll@med.uni-jena.de Klinikum der Friedrich-Schiller-Universität Jena |
| Competitive hybridization on glass DNA-micro-arrays is the most widely used method for gene expression profiling in biological samples. Unfortunately, due to accumulation of variations occurring at the different experimental steps, the data show significant differences that do not reflect the true expression profile of the biological sample. Because of nonlinear effects simple normalization methods (e.g. reference genes or globalization) which are based on linear scaling, will fail to normalize the data. Therefore methods like lowess (local regression) [1] were applied. But especially lowess tends to process the data in a "blind" fashion. This can lead to overfitting and finally to normalization artifacts. We assume that most of the nonlinear effects come from nonlinear signal functions in the fluorescence measurement. This can be corrected with channel-based normalization. For this we combined our ranking based normalization [2] with a 2nd order polynomial fit between the two channels of the competitive hybridization. The method was tested on several data sets from independent hybridization experiments. |
| [1] Yang YH et al. Nucleic Acids Res 2002 Volume 30, e15 [2] Kroll TC et al. Nucleic Acids Res 2002 Volume 30, e50 |