We present a fast, versatile and adaptive-multiscale algorithm for analyzing a wide-variety of DNA microarray data. Its primary application is in normalization of array data as well as subsequent identification of 'enriched targets', e.g. differentially expressed genes in expression profiling arrays and enriched sites in ChIP-on-chip experimental data. We show how to accommodate the unique characteristics of ChIP-on-chip data, where the set of 'enriched targets' is large, asymmetric and whose proportion to the whole data varies locally.
Bibliographical noteFunding Information:
The authors are grateful to the NYU Cancer Institute Genomics Facility for providing necessary instrumentation and expertise. The authors thank Mary Tsikitis and Diego Acosta for their help in performing confirmatory PCR, and Rick Young and Duncan Odom of the Whitehead Institute for advice in the design of a mouse promoter Microarray. The authors also thank Fang Cheng, Ronald R. Coifman, Peter Jones and Yi (Joey) Zhou for helpful discussions; E. Terry Papoutsakis and Carles Paredes for their help in interpreting the data appearing in their original paper on pSOL1 genes; James Glimm and Jacob Schwartz for commenting on earlier versions of this paper and finally, Mark Green and IPAM (UCLA) for inviting G.L. and J.M. to take part in their bioinfor-matics as well as multiscale geometry meetings, where discussions of similar topics stimulated our research. Special thanks for the careful anonymous reviewers and their constructive suggestions. J.M., G.L. and B.M. are supported by grants from NSF’s ITR program, Defense Advanced Research Projects Agency (DARPA), and New York State Office of Science, Technology & Academic Research (NYSTAR). G.L. is supported by NSF grant #0612608.
B.D. is supported by an NIH grant #5R01 GM067132-02. A.B. is supported by a postdoctoral training fellowship from the Fonds de la Recherche en Sante du Quebec. Funding to pay the Open Access publication charges for this article was provided by NSF grant #0612608.