Open AccessPost‐normalization quality assessment visualization of microarray data
Author(s) -
McClure John,
Wit Ernst
Publication year - 2003
Publication title -
comparative and functional genomics
Language(s) - English
Resource type - Journals
eISSN - 1532-6268
pISSN - 1531-6912
DOI - 10.1002/cfg.317
Subject(s) - normalization (sociology) , computer science , database normalization , data mining , visualization , inference , data quality , dna microarray , artificial intelligence , pattern recognition (psychology) , engineering , biology , metric (unit) , biochemistry , operations management , gene expression , sociology , anthropology , gene
Post‐normalization checking of microarrays rarely occurs, despite the problems that using unreliable data for inference can cause. This paper considers a number of different ways to check microarrays after normalization for a variety of potential problems. Four types of problem with microarray data that these checks can identify are: clerical mistakes, array‐wide hybridization problems, problems with normalization and mishandling problems. Any of these can seriously affect the results of any analysis. The three main techniques used to identify these problems are dimension reduction techniques, false array plots and correlograms. None of the techniques are computationally very intensive and all can be carried out in the R statistical package. Once discovered, problems can either be rectified or excluded from the data. Copyright © 2003 John Wiley & Sons, Ltd.